Files
edx-platform/common/djangoapps/track/shim.py
brianhw b3b3ef98b2 feat: add courserun_key property to GA event transformer. (#28761)
The GA transformer adds the context.course_id property of a tracking
log event to the event transmitted automatically to Segment in the
"label" property, so GA can use it.  This adds it also to the
"courserun_key" property, so that regular users of Segment events can
also query it in a more consistently-named property.  However, it does
so only if it is a valid CourseKey.

For DENG-997.

Add CourseKey parsing.
2021-10-19 16:40:30 -04:00

123 lines
4.0 KiB
Python

"""Map new event context values to old top-level field values. Ensures events can be parsed by legacy parsers."""
import json
from opaque_keys import InvalidKeyError
from opaque_keys.edx.keys import CourseKey
from .transformers import EventTransformerRegistry
CONTEXT_FIELDS_TO_INCLUDE = [
'username',
'session',
'ip',
'agent',
'host',
'referer',
'accept_language'
]
class LegacyFieldMappingProcessor:
"""Ensures all required fields are included in emitted events"""
def __call__(self, event):
context = event.get('context', {})
if 'context' in event:
for field in CONTEXT_FIELDS_TO_INCLUDE:
self.move_from_context(field, event)
remove_shim_context(event)
if 'data' in event:
if context.get('event_source', '') == 'browser' and isinstance(event['data'], dict):
event['event'] = json.dumps(event['data'])
else:
event['event'] = event['data']
del event['data']
else:
event['event'] = {}
if 'timestamp' in context:
event['time'] = context['timestamp']
del context['timestamp']
elif 'timestamp' in event:
event['time'] = event['timestamp']
if 'timestamp' in event:
del event['timestamp']
self.move_from_context('event_type', event, event.get('name', ''))
self.move_from_context('event_source', event, 'server')
self.move_from_context('page', event, None)
def move_from_context(self, field, event, default_value=''):
"""Move a field from the context to the top level of the event."""
context = event.get('context', {})
if field in context:
event[field] = context[field]
del context[field]
else:
event[field] = default_value
def remove_shim_context(event):
"""
Remove obsolete fields from event context.
"""
if 'context' in event:
context = event['context']
# These fields are present elsewhere in the event at this point
context_fields_to_remove = set(CONTEXT_FIELDS_TO_INCLUDE)
# This field is only used for Segment web analytics and does not concern researchers
context_fields_to_remove.add('client_id')
for field in context_fields_to_remove:
if field in context:
del context[field]
class GoogleAnalyticsProcessor:
"""Adds course_id as label, and sets nonInteraction property"""
# documentation of fields here: https://segment.com/docs/integrations/google-analytics/
# this should *only* be used on events destined for segment.com and eventually google analytics
def __call__(self, event):
context = event.get('context', {})
course_id = context.get('course_id')
copied_event = event.copy()
if course_id is not None:
copied_event['label'] = course_id
# The value stored as course_id is not always a courserun_key.
# It may, for example, be a library instead. So we parse it first to be sure.
# We add a str() call to the input so that sentinel values don't cause
# CourseKey to spit up with a different error.
try:
courserun_key = CourseKey.from_string(str(course_id))
copied_event['courserun_key'] = str(courserun_key)
except InvalidKeyError:
pass
copied_event['nonInteraction'] = 1
return copied_event
class PrefixedEventProcessor:
"""
Process any events whose name or prefix (ending with a '.') is registered
as an EventTransformer.
"""
def __call__(self, event):
"""
If the event is registered with the EventTransformerRegistry, transform
it. Otherwise do nothing to it, and continue processing.
"""
try:
event = EventTransformerRegistry.create_transformer(event)
except KeyError:
return
event.transform()
return event