Django_Tables2 Dynamic Columns with Filters - django

Question

Goal was to implement a simple View for the Users to Select columns dynamically with some added calculated info (annotations on some specific columns) and also let them filter on fields.
Thankful for any comments, since this took me quite a few hours to get it working properly I thought I would provide a short writeup for anyone looking at a similar problem :)
Used Modules/Libraries etc:
Django-Filter
Django_Tables2
Bootstrap-Select to properly display Multiple Choice Fields
Example Model which we would like to use:
class Summary(models.Model):
billing_date = models.DateField(verbose_name='Billing Date')
period = models.CharField(max_length=10, verbose_name='Period')
operator = models.CharField(max_length=50, verbose_name='Operator')
product = models.CharField(max_length=30, verbose_name='Product')
...

Answer 1

The filters are really straightforward, the only special case here is that the we want an empty queryset initially and some fields should be required.
"info" will hold the select columns of our "Summary" Model, "product" and "operator" are just fields in Summary.
class AdHocReportFilter(django_filters.FilterSet):
info = django_filters.MultipleChoiceFilter(choices=report_field_choices, label='Available Fields', required=True)
product = django_filters.ModelChoiceFilter(queryset=Product.objects.all(), label='Products', required=True)
operator = django_filters.CharFilter(field_name="operator", lookup_expr='contains', label='Operator')
....
def __init__(self, *args, **kwargs):
super(AdHocReportFilter, self).__init__(*args, **kwargs)
if self.data == {}:
self.queryset = self.queryset.none()
Template:
Nothing interesting to show here, you can use Bootstrap-Select to tidy up your Multi Select Fields (there are quite a few nice writeups about that available).
Make sure to put your Table into an "if" as the object may or may not exist depending on your view (if someone wants an example of the template let me know)
View:
Extract your GET requests accordingly (either as list or simple value depending on your available filters).
The actual filter itself depends on what you want the user to be able to filter, make sure to either make all necessary fields required or replace them with some standard value as the filter will not accept None Types.
"field__contains" is your friend here since it will also show values on not selected fields!
Special Case, if there can actually be "Null" Values in the DB for specific fields, move them to another filter likethe below example of the Q - query!
Fortunately "values" accepts "*list" which is a simple list of all our available columns.
The annotations are just dependant on what you want to achieve.
Call the Table Object with the added argument "user_columns" which holds our list so we can build the required Table.
#login_required
def ad_hoc_report(request):
template_name = 'non_voice/ad_hoc_report.html'
filter = AdHocReportFilter(request.GET, queryset=Summary.objects.all())
info = request.GET.getlist('info', None)
product = request.GET.get('product', '')
operator = request.GET.get('operator', '')
start_date = request.GET.get('start_date', None)
end_date = request.GET.get('end_date', None)
if operator is None:
operator = ''
result_object = Summary.objects.filter(product__contains=product, ).filter((Q(operator__contains=operator)|Q(operator__isnull=True)).values(*info).annotate(
Amount=Sum("amount"), Count=Sum("count"))
table = AdHocReportTable(data=result_object, user_columns=info)
return render(request, template_name, {'filter': filter, 'table': table})
Table:
This was the difficult part and only possible with lots and lots of reading various stack overflow comments :)
First of all define your calculated annotation columns and set the required Meta info, the '...' is a built in placeholder without knowing the column name in advance (which helps us to move our calculated columns to the end of the Table)
In the init we first check if our "self.base_columns" are consistent with what we provided and remove columns which were deselected by our user, otherwise it would still show them empty even after filtering. (Maybe there is a nicer way to do this, haven't found it yet)
In the next step add the columns selected by our user dynamically from the mentioned above "user_columns" which we passed in the views.py
class AdHocReportTable(tables.Table):
Amount = tables.Column(verbose_name='Amount')
Count = tables.Column(verbose_name='Count')
class Meta:
# '...' is a built in placeholder!
sequence = ('...', 'Amount', 'Count')
template_name = "django_tables2/bootstrap4.html"
attrs = {'class': 'table table-hover', }
# This makes it possible to pass a dynamic list of columns to the Table Object
def __init__(self, data, user_columns, *args, **kwargs):
if user_columns:
calulated_columns = ['Amount', 'Count']
# Removes deselected columns from the table (otherwise they are shown empty)
for key, val in self.base_columns.items():
if key not in user_columns and key not in calulated_columns:
del self.base_columns[key]
# Add the Selected Columns dynamically to the Table
for col in user_columns:
self.base_columns[col] = tables.Column(verbose_name=col)
super(AdHocReportTable, self).__init__(data, user_columns, *args, **kwargs)