Using a renderer in DRF settings for Excel output option about django-rest-pandas

[sheppard]

This use case wasn't fully supported before, but it was an easy fix. I added some more descriptive error messages and also updated the documentation here. The error is that if you have a rest_pandas renderer in DEFAULT_RENDERER_CLASSES, you need to import rest_pandas before importing anything from rest_framework.views or you may end up with a recursive import.

One way to fix the import order is to add rest_pandas to your INSTALLED_APPS, though you'll also want to import PandasMixin to add to your own views.

[sssolid]

Response data is a OrderedDict, not a DataFrame! Did you extend PandasMixin?

In my case, the above error was directly related to pagination and I didn't want to disable DEFAULT_PAGINATION_CLASS. As a quick workaround for the problem, I override ListMixin and check to see if the URL parameter "format" exists. I don't use a format extension (/path.csv) so I didn't bother to add it.

from django.conf import settings
from rest_framework import mixins

class PaginatedPandaListMixin(mixins.ListModelMixin):

    def list(self, request, *args, **kwargs):
        queryset = self.filter_queryset(self.get_queryset())
        # Check if format parameter overridden
        try:
            format_param = settings.URL_FORMAT_OVERRIDE
        except AttributeError:
            format_param = 'format'
        # Don't paginate if response is for DRP or API
        if request.GET.get(format_param) and not request.GET.get(format_param) == 'api':
            serializer = self.get_serializer(queryset, many=True)
            return Response(serializer.data)
        page = self.paginate_queryset(queryset)
        if page is not None:
            serializer = self.get_serializer(page, many=True)
            return self.get_paginated_response(serializer.data)
        serializer = self.get_serializer(queryset, many=True)
        return Response(serializer.data)

I then use the above mixin on my viewsets and all is well.

class CustomViewSet(PandasMixin, PaginatedPandaListMixin, ModelViewSet):

[gatensj]

Hey there. Thanks for the response. I've followed your instructions listed above, and am still running into issues. In my settings file, I've included rest_pandas in my included apps. Then in my models, I defined a objects = DataFrameManager(), right before the class Meta definition. My serializer is a ModelSerializer. And my View is passing PandasMixin, ListAPIView. Listed below is my code.

`

(serializer.py)

    class FfAllFactorsDailyPandaSerializer(ModelSerializer):
            class Meta:
                model = FfAllFactorsDailyModel
                fields = (
                    'date',
                    'mktrf',
                    'smb',
                    'hml',
                    'rf',
                    'umd',
                )

(views.py)

    class FfAllSeriesView(PandasMixin, ListAPIView):
            queryset = FfAllFactorsDailyModel.objects.all()
            serializer_class = FfAllFactorsDailyPandaSerializer
            permission_classes = (FfAllPermission,)

            def filter_queryset(self, qs):
                # At this point, you can filter queryset based on self.request or other
                # settings (useful for limiting memory usage).  This function can be
                # omitted if you are using a filter backend or do not need filtering.
                return qs

`

Much to my chagrin, I'm still getting a "Response data is a <class 'rest_framework.utils.serializer_helpers.ReturnList'>, not a DataFrame!" error. In short, what am I doing wrong? I thought the (objects = DataFrameManager()) defined in the model, would return a DataFrame. Thanks in advance for the assist. Take care and have a good day.

Sincerely,

JG

[FlipperPA]

Thanks very much for our efforts, @sheppard - but I'm having a similar issue to JG, and I think I'm close. I've installed the latest rest_pandas from master. I've added rest_pandas to my INSTALLED_APPS. Here's are my settings:

REST_FRAMEWORK = {
    'DEFAULT_PERMISSION_CLASSES': (
        'rest_framework.permissions.IsAuthenticated',
    ),
    'DEFAULT_FILTER_BACKENDS': (
        'django_filters.rest_framework.DjangoFilterBackend',
    ),
    'DEFAULT_AUTHENTICATION_CLASSES': (
        'rest_framework.authentication.BasicAuthentication',
        'rest_framework.authentication.SessionAuthentication',
        'rest_framework.authentication.TokenAuthentication',
    ),
    'DEFAULT_METADATA_CLASS': 'home.adapters.TypesFiltersMetadata',
    'DEFAULT_PAGINATION_CLASS': 'rest_framework.pagination.LimitOffsetPagination',
    'PAGE_SIZE': 10,

    'DEFAULT_RENDERER_CLASSES': (
        'rest_framework.renderers.JSONRenderer',
        'rest_framework.renderers.BrowsableAPIRenderer',
        'rest_pandas.renderers.PandasExcelRenderer',
    ),
}

I've modified my ViewSets to look like this:

class MyExampleViewSet(PandasMixin, ReadOnlyModelViewSet):
    queryset = MyExampleModel.objects.all()
    serializer_class = MyExampleSerializer
    permission_classes = (MyPermission,)
    filter_backends = (DjangoFilterBackend,)
    filter_fields = (
        'date',
    )

I can browse the HTML version of the DRF API, and xlsx appears as an option in the drop down alongside json and html. However, when I select xlsx I'm getting this error, despite having extended PandasMixin:

Response data is a OrderedDict, not a DataFrame! Did you extend PandasMixin?

I've started digging through the code, but any help you could provide would be fantastic. Thanks again for your efforts on DRP - much appreciated!

[mehta]

Remove DEFAULT_PAGINATION_CLASS in your settings.