edx-enterprise-data 9.0.0__tar.gz → 9.1.0__tar.gz

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0}/CHANGELOG.rst

@@ -15,9 +15,14 @@ Unreleased
 ----------
 
 =========================
-[9.0.0] - 2024-09-23
+
+[9.1.0] - 2024-09-23
+---------------------
+  * refactor: Performance optimizations for engagement and completions related API endpoints.
+
+[9.0.1] - 2024-09-23
 ---------------------
-  * feat!: Python 3.12 Upgrade. Dropped support for ``Python<3.12``.
+  * revert: Revert "feat!: Python 3.12 Upgrade"
 
 [8.13.0] - 2024-09-23
 ---------------------

{edx_enterprise_data-9.0.0/edx_enterprise_data.egg-info → edx_enterprise_data-9.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: edx-enterprise-data
-Version: 9.0.0
+Version: 9.1.0
 Summary: Enterprise Reporting
 Home-page: https://github.com/openedx/edx-enterprise-data
 Author: edX
@@ -9,6 +9,7 @@ License: AGPL 3.0
 Classifier: Framework :: Django
 Classifier: Framework :: Django :: 4.2
 Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.12
 License-File: LICENSE
 Requires-Dist: Django
@@ -23,8 +24,8 @@ Requires-Dist: edx-rbac
 Requires-Dist: edx-rest-api-client
 Requires-Dist: factory_boy
 Requires-Dist: mysql-connector-python
-Requires-Dist: numpy
-Requires-Dist: pandas
+Requires-Dist: numpy<=1.24.4
+Requires-Dist: pandas<=2.0.3
 Requires-Dist: requests
 Requires-Dist: rules
 Provides-Extra: reporting

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0/edx_enterprise_data.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: edx-enterprise-data
-Version: 9.0.0
+Version: 9.1.0
 Summary: Enterprise Reporting
 Home-page: https://github.com/openedx/edx-enterprise-data
 Author: edX
@@ -9,6 +9,7 @@ License: AGPL 3.0
 Classifier: Framework :: Django
 Classifier: Framework :: Django :: 4.2
 Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.12
 License-File: LICENSE
 Requires-Dist: Django
@@ -23,8 +24,8 @@ Requires-Dist: edx-rbac
 Requires-Dist: edx-rest-api-client
 Requires-Dist: factory_boy
 Requires-Dist: mysql-connector-python
-Requires-Dist: numpy
-Requires-Dist: pandas
+Requires-Dist: numpy<=1.24.4
+Requires-Dist: pandas<=2.0.3
 Requires-Dist: requests
 Requires-Dist: rules
 Provides-Extra: reporting

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0}/edx_enterprise_data.egg-info/SOURCES.txt

@@ -21,7 +21,6 @@ enterprise_data/signals.py
 enterprise_data/urls.py
 enterprise_data/utils.py
 enterprise_data/admin_analytics/__init__.py
-enterprise_data/admin_analytics/completions_utils.py
 enterprise_data/admin_analytics/constants.py
 enterprise_data/admin_analytics/data_loaders.py
 enterprise_data/admin_analytics/utils.py
@@ -47,12 +46,12 @@ enterprise_data/api/v1/paginators.py
 enterprise_data/api/v1/serializers.py
 enterprise_data/api/v1/urls.py
 enterprise_data/api/v1/views/__init__.py
+enterprise_data/api/v1/views/analytics_completions.py
 enterprise_data/api/v1/views/analytics_engagements.py
 enterprise_data/api/v1/views/analytics_enrollments.py
 enterprise_data/api/v1/views/analytics_leaderboard.py
 enterprise_data/api/v1/views/base.py
 enterprise_data/api/v1/views/enterprise_admin.py
-enterprise_data/api/v1/views/enterprise_completions.py
 enterprise_data/api/v1/views/enterprise_learner.py
 enterprise_data/api/v1/views/enterprise_offers.py
 enterprise_data/fixtures/enterprise_enrollment.json

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0}/edx_enterprise_data.egg-info/requires.txt

@@ -10,8 +10,8 @@ edx-rbac
 edx-rest-api-client
 factory_boy
 mysql-connector-python
-numpy
-pandas
+numpy<=1.24.4
+pandas<=2.0.3
 requests
 rules
 

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0}/enterprise_data/__init__.py

@@ -2,4 +2,4 @@
 Enterprise data api application. This Django app exposes API endpoints used by enterprises.
 """
 
-__version__ = "9.0.0"
+__version__ = "9.1.0"

edx_enterprise_data-9.1.0/enterprise_data/admin_analytics/constants.py

@@ -0,0 +1,25 @@
+"""Advanced Analytics Constants"""
+
+from enum import Enum
+
+
+class Granularity(Enum):
+    """Granularity choices"""
+    DAILY = 'Daily'
+    WEEKLY = 'Weekly'
+    MONTHLY = 'Monthly'
+    QUARTERLY = 'Quarterly'
+
+
+class Calculation(Enum):
+    """Calculation choices"""
+    TOTAL = 'Total'
+    RUNNING_TOTAL = 'Running Total'
+    MOVING_AVERAGE_3_PERIOD = 'Moving Average (3 Period)'
+    MOVING_AVERAGE_7_PERIOD = 'Moving Average (7 Period)'
+
+
+class ResponseType(Enum):
+    """Response type choices"""
+    JSON = 'json'
+    CSV = 'csv'

edx_enterprise_data-9.1.0/enterprise_data/admin_analytics/database/queries/fact_engagement_admin_dash.py

@@ -0,0 +1,111 @@
+"""
+Module containing queries for the fact_enrollment_engagement_day_admin_dash table.
+"""
+
+
+class FactEngagementAdminDashQueries:
+    """
+    Queries related to the fact_enrollment_engagement_day_admin_dash table.
+    """
+    @staticmethod
+    def get_learning_hours_and_daily_sessions_query():
+        """
+        Get the query to fetch the learning hours and daily sessions.
+        """
+        return """
+            SELECT
+                ROUND(SUM(learning_time_seconds) / 60 / 60, 1) as hours, SUM(is_engaged) as sessions
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                activity_date BETWEEN %(start_date)s AND %(end_date)s;
+        """
+
+    @staticmethod
+    def get_engagement_count_query():
+        """
+        Get the query to fetch the total number of engagements for an enterprise customer.
+        """
+        return """
+            SELECT count(*)
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                activity_date BETWEEN %(start_date)s AND %(end_date)s;
+        """
+
+    @staticmethod
+    def get_all_engagement_query():
+        """
+        Get the query to fetch all engagement data.
+        """
+        return """
+            SELECT
+                email, course_title, course_subject, enroll_type, activity_date,
+                learning_time_seconds/3600 as learning_time_hours
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                activity_date BETWEEN %(start_date)s AND %(end_date)s
+            ORDER BY activity_date DESC LIMIT %(limit)s OFFSET %(offset)s;
+        """
+
+    @staticmethod
+    def get_top_courses_by_engagement_query(record_count=20):
+        """
+        Get the query to fetch the learning time in hours by courses.
+
+        Query should fetch the top N courses by user engagement. Where N is the value of record_count.
+
+        Arguments:
+            record_count (int): Number of records to fetch.
+
+        Returns:
+            (str): Query to fetch the learning time in hours by courses for the top courses by user engagement.
+        """
+        return f"""
+            SELECT course_key, course_title, enroll_type, SUM(learning_time_seconds)/3600 as learning_time_hours
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                activity_date BETWEEN %(start_date)s AND %(end_date)s
+            GROUP BY course_key, course_title, enroll_type
+            ORDER BY learning_time_hours DESC LIMIT {record_count};
+        """
+
+    @staticmethod
+    def get_top_subjects_by_engagement_query(record_count=20):
+        """
+        Get the query to fetch the learning time in hours by subjects.
+
+        Query should fetch the top N subjects by user engagement. Where N is the value of record_count.
+
+        Arguments:
+            record_count (int): Number of records to fetch.
+
+        Returns:
+            (str): Query to fetch the learning time in hours by subjects for the top subjects by user engagement.
+        """
+        return f"""
+            SELECT course_subject, enroll_type, SUM(learning_time_seconds)/3600 as learning_time_hours
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                activity_date BETWEEN %(start_date)s AND %(end_date)s
+            GROUP BY course_subject, enroll_type
+            ORDER BY learning_time_hours DESC LIMIT {record_count};
+        """
+
+    @staticmethod
+    def get_engagement_time_series_data_query():
+        """
+        Get the query to fetch the completion time series data.
+
+        Query should fetch the time series data with daily granularity.
+
+        Returns:
+            (str): Query to fetch the completion time series data.
+        """
+        return """
+            SELECT activity_date, enroll_type, SUM(learning_time_seconds)/3600 as learning_time_hours
+            FROM fact_enrollment_engagement_day_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                activity_date BETWEEN %(start_date)s AND %(end_date)s
+            GROUP BY activity_date, enroll_type
+            ORDER BY activity_date;
+        """

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0}/enterprise_data/admin_analytics/database/queries/fact_enrollment_admin_dash.py

@@ -68,6 +68,7 @@ class FactEnrollmentAdminDashQueries:
                 SUM(has_passed) as completions
             FROM fact_enrollment_admin_dash
             WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                has_passed=1 AND
                 passed_date BETWEEN %(start_date)s AND %(end_date)s;
         """
 
@@ -85,7 +86,7 @@ class FactEnrollmentAdminDashQueries:
         """
 
     @staticmethod
-    def get_top_courses_enrollments_query(record_count=20):
+    def get_top_courses_by_enrollments_query(record_count=20):
         """
         Get the query to fetch the enrollment count by courses.
 
@@ -94,14 +95,12 @@ class FactEnrollmentAdminDashQueries:
         Arguments:
             record_count (int): Number of records to fetch.
         """
-        # Some local environments raise error when course_title is added in SELECT without GROUP BY.
-        # If you face this issue, you can remove course_title from SELECT.
         return f"""
             SELECT course_key, course_title , enroll_type, count(course_key) as enrollment_count
             FROM fact_enrollment_admin_dash
             WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
                 enterprise_enrollment_date BETWEEN %(start_date)s AND %(end_date)s
-            GROUP BY course_key, enroll_type ORDER BY enrollment_count DESC LIMIT {record_count};
+            GROUP BY course_key, course_title, enroll_type ORDER BY enrollment_count DESC LIMIT {record_count};
         """
 
     @staticmethod
@@ -135,3 +134,83 @@ class FactEnrollmentAdminDashQueries:
             GROUP BY enterprise_enrollment_date, enroll_type
             ORDER BY enterprise_enrollment_date;
         """
+
+    @staticmethod
+    def get_all_completions_query():
+        """
+        Get the query to fetch all completions.
+        """
+        return """
+            SELECT email, course_title, course_subject, enroll_type, passed_date
+            FROM fact_enrollment_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                has_passed=1 AND
+                passed_date BETWEEN %(start_date)s AND %(end_date)s
+            ORDER BY passed_date DESC LIMIT %(limit)s OFFSET %(offset)s
+        """
+
+    @staticmethod
+    def get_top_courses_by_completions_query(record_count=20):
+        """
+        Get the query to fetch the completion count by courses.
+
+        Query should fetch the top N courses by completion count. Where N is the value of record_count.
+
+        Arguments:
+            record_count (int): Number of records to fetch.
+
+        Returns:
+            (str): Query to fetch the enrollment count by courses for the top courses by enrollment count.
+        """
+        return f"""
+            SELECT course_key, course_title, enroll_type, count(course_key) as completion_count
+            FROM fact_enrollment_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                has_passed=1 AND
+                passed_date BETWEEN %(start_date)s AND %(end_date)s
+            GROUP BY course_key, course_title, enroll_type
+            ORDER BY completion_count DESC LIMIT {record_count};
+        """
+
+    @staticmethod
+    def get_top_subjects_by_completions_query(record_count=20):
+        """
+        Get the query to fetch the completion count by subjects.
+
+        Query should fetch the top N subjects by completion count. Where N is the value of record_count.
+
+        Arguments:
+            record_count (int): Number of records to fetch.
+
+        Returns:
+            (str): Query to fetch the completion count by subjects for the top subjects by completion count.
+        """
+        return f"""
+            SELECT course_subject, enroll_type, count(course_subject) as completion_count
+            FROM fact_enrollment_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                has_passed=1 AND
+                passed_date BETWEEN %(start_date)s AND %(end_date)s
+            GROUP BY course_subject, enroll_type
+            ORDER BY completion_count DESC LIMIT {record_count};
+        """
+
+    @staticmethod
+    def get_completions_time_series_data_query():
+        """
+        Get the query to fetch the completion time series data.
+
+        Query should fetch the time series data with daily granularity.
+
+        Returns:
+            (str): Query to fetch the completion time series data.
+        """
+        return """
+            SELECT passed_date, enroll_type, count(course_key) as completion_count
+            FROM fact_enrollment_admin_dash
+            WHERE enterprise_customer_uuid=%(enterprise_customer_uuid)s AND
+                has_passed=1 AND
+                passed_date BETWEEN %(start_date)s AND %(end_date)s
+            GROUP BY passed_date, enroll_type
+            ORDER BY passed_date;
+        """

edx_enterprise_data-9.1.0/enterprise_data/admin_analytics/database/tables/fact_engagement_admin_dash.py

@@ -0,0 +1,154 @@
+"""
+Module for interacting with the fact_enrollment_engagement_day_admin_dash table.
+"""
+from datetime import date
+from uuid import UUID
+
+from ..queries import FactEngagementAdminDashQueries
+from ..utils import run_query
+from .base import BaseTable
+
+
+class FactEngagementAdminDashTable(BaseTable):
+    """
+    Class for communicating with the fact_enrollment_engagement_day_admin_dash table.
+    """
+    queries = FactEngagementAdminDashQueries()
+
+    def get_learning_hours_and_daily_sessions(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the learning hours and daily sessions for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            (tuple<float, int>): The learning hours and daily sessions.
+        """
+        results = run_query(
+            query=self.queries.get_learning_hours_and_daily_sessions_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            }
+        )
+        return tuple(results[0])
+
+    def get_engagement_count(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the total number of engagements for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            int: The total number of engagements.
+        """
+        results = run_query(
+            query=self.queries.get_engagement_count_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            }
+        )
+        return results[0][0]
+
+    def get_all_engagements(
+            self, enterprise_customer_uuid: UUID, start_date: date, end_date: date, limit: int, offset: int
+    ):
+        """
+        Get all engagement data for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+            limit (int): The maximum number of records to return.
+            offset (int): The number of records to skip.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the engagement data.
+        """
+        return run_query(
+            query=self.queries.get_all_engagement_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+                'limit': limit,
+                'offset': offset,
+            },
+            as_dict=True,
+        )
+
+    def get_top_courses_by_engagement(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the top courses by user engagement for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the course data.
+        """
+        return run_query(
+            query=self.queries.get_top_courses_by_engagement_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )
+
+    def get_top_subjects_by_engagement(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the top subjects by user engagement for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the subject data.
+        """
+        return run_query(
+            query=self.queries.get_top_subjects_by_engagement_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )
+
+    def get_engagement_time_series_data(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the engagement time series data.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the engagement time series data.
+        """
+        return run_query(
+            query=self.queries.get_engagement_time_series_data_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )

{edx_enterprise_data-9.0.0 → edx_enterprise_data-9.1.0}/enterprise_data/admin_analytics/database/tables/fact_enrollment_admin_dash.py

@@ -35,7 +35,7 @@ class FactEnrollmentAdminDashTable(BaseTable):
                 'end_date': end_date,
             }
         )
-        return results[0][0]
+        return int(results[0][0] or 0)
 
     def get_all_enrollments(
             self, enterprise_customer_uuid: UUID, start_date: date, end_date: date, limit: int, offset: int
@@ -132,7 +132,7 @@ class FactEnrollmentAdminDashTable(BaseTable):
                 'end_date': end_date,
             }
         )
-        return results[0][0]
+        return int(results[0][0] or 0)
 
     def get_top_courses_by_enrollments(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
         """
@@ -147,7 +147,7 @@ class FactEnrollmentAdminDashTable(BaseTable):
             list<dict>: A list of dictionaries containing the course key, course_title and enrollment count.
         """
         return run_query(
-            query=self.queries.get_top_courses_enrollments_query(),
+            query=self.queries.get_top_courses_by_enrollments_query(),
             params={
                 'enterprise_customer_uuid': enterprise_customer_uuid,
                 'start_date': start_date,
@@ -199,3 +199,97 @@ class FactEnrollmentAdminDashTable(BaseTable):
             },
             as_dict=True,
         )
+
+    def get_all_completions(
+            self, enterprise_customer_uuid: UUID, start_date: date, end_date: date, limit: int, offset: int
+    ):
+        """
+        Get all completions for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+            limit (int): The maximum number of records to return.
+            offset (int): The number of records to skip.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the completions data.
+        """
+        return run_query(
+            query=self.queries.get_all_completions_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+                'limit': limit,
+                'offset': offset,
+            },
+            as_dict=True,
+        )
+
+    def get_top_courses_by_completions(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the top courses by completion for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the course key, course_title and completion count.
+        """
+        return run_query(
+            query=self.queries.get_top_courses_by_completions_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )
+
+    def get_top_subjects_by_completions(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the top subjects by completions for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the subject and completion count.
+        """
+        return run_query(
+            query=self.queries.get_top_subjects_by_completions_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )
+
+    def get_completions_time_series_data(self, enterprise_customer_uuid: UUID, start_date: date, end_date: date):
+        """
+        Get the completions time series data for the given enterprise customer.
+
+        Arguments:
+            enterprise_customer_uuid (UUID): The UUID of the enterprise customer.
+            start_date (date): The start date.
+            end_date (date): The end date.
+
+        Returns:
+            list<dict>: A list of dictionaries containing the date and completion count.
+        """
+        return run_query(
+            query=self.queries.get_completions_time_series_data_query(),
+            params={
+                'enterprise_customer_uuid': enterprise_customer_uuid,
+                'start_date': start_date,
+                'end_date': end_date,
+            },
+            as_dict=True,
+        )