Limit resources access view to show only events for enrolled students…

… in the course #764 (#1460)

Co-authored-by: Sam Sciolla <35741256+ssciolla@users.noreply.github.com>

dashboard/cron.py

@@ -17,7 +17,7 @@
 from sqlalchemy.engine import ResultProxy
 
 from dashboard.common import db_util, utils
-from dashboard.models import Course, Resource, AcademicTerms, ResourceAccess
+from dashboard.models import Course, Resource, AcademicTerms, User
 
 
 logger = logging.getLogger(__name__)
@@ -343,7 +343,7 @@ def update_resource_access(self):
                     resource_access_df = resource_access_df \
                         .drop(columns=['user_id_str', 'user_login_name']) \
                         .dropna()
-
+                    resource_access_df['user_id'] = pd.to_numeric(resource_access_df['user_id'])
                     logger.debug(f'resource_access_df:\n'
                                  f'{resource_access_df}\n'
                                  f'{resource_access_df.dtypes}')
@@ -389,7 +389,17 @@ def update_resource_access(self):
             logger.debug(f'resource_access_df:\n'
                          f'{resource_access_df}\n'
                          f'{resource_access_df.dtypes}')
-
+            # only keep access events generated by students
+            student_enrollment_type = User.EnrollmentType.STUDENT
+            student_enrollment_df = pd.read_sql(
+                'select user_id, course_id from user where enrollment_type= %s',
+                engine, params={student_enrollment_type})
+            resource_access_df = pd.merge(
+                resource_access_df, student_enrollment_df,
+                on=['user_id', 'course_id'],
+                # use inner merge to keep only resource access event (left)
+                # innitiated by people with student enrollment type (right)
+                how='inner')
             # First, update resource table
             try:
                 dtype = {'resource_id': types.VARCHAR(255)}