Superevents

Adding superevents to GraceDB, including the necessary models, utility functions, and API resources. This has also resulted in changes to the overall configuration and some parts of the events model.

Many models in the events app, like Labels, Tags, EventLogs, etc. were needed in a similar form for superevents. We were able to completely reuse the existing Label and Tag models for superevents. In other cases, we had to abstract the main functionality of the class into an abstract base class and then implement separate models for events and superevents, due to foreign key requirements. This should leave the current functionality of the events models unchanged.

Some portions of the events API have been modified because of one of the following reasons:

• It didn't make sense and was unnecessarily difficult to duplicate for superevents
• It was inefficient or bad

These changes are documented here: https://bugs.ligo.org/redmine/issues/6133

gracedb/superevents/api/fields.py

 from rest_framework import fields
 
+import six
+import logging
+logger = logging.getLogger(__name__)
+
+
 class CustomHiddenDefault(fields.CurrentUserDefault):
     context_key = None
 
@@ -37,3 +42,25 @@ class ParentObjectDefault(CustomHiddenDefault):
                 value = value()
             #value = serializer_field.context['view'].get_parent()
         return value
+
+
+class CommaSeparatedOrListField(fields.ListField):
+    default_style = {'base_template': 'input.html'}
+
+    def __init__(self, *args, **kwargs):
+        super(CommaSeparatedOrListField, self).__init__(*args, **kwargs)
+        # Set form style for browsable API
+        self.style = kwargs.get('style', self.default_style)
+
+    def to_internal_value(self, data):
+        # Empirical tests with HTML forms indicate that if we enter
+        # something like 1,2,3 in a form, we will get something like
+        # [u'1,2,3'] here.  So if we get input like that, we convert it
+        # to [u'1', u'2', u'3'], then pass it to the base class's
+        # to_internal_value() method. Might not be safe for cases where
+        # a list contains CharFields which might have commas in them.
+        if (isinstance(data, list) and len(data) == 1 and
+            isinstance(data[0], six.string_types)):
+            data = data[0].split(',')
+        return super(CommaSeparatedOrListField, self).to_internal_value(data)
+