fix: use final mv and add total video count

models/grading/fact_student_status.sql

@@ -13,7 +13,7 @@ select
     users.name as name,
     users.email as email,
     fes.emission_time as enrolled_at
-from {{ ref("fact_enrollment_status") }} fes
+from {{ ref("fact_enrollment_status") }} fes FINAL
 left join
     {{ ref("fact_learner_course_status") }} lg
     on fes.org = lg.org

models/grading/unit_tests.yaml

@@ -38,33 +38,3 @@ unit_tests:
     expect:
       format: csv
       fixture: fact_learner_course_status_expected
-
-  - name: test_fact_student_status
-    model: fact_student_status
-    config:
-      tags: 'ci'
-    given:
-      - input: ref('fact_enrollment_status')
-        format: sql
-        rows: |
-          select * from fact_enrollment_status
-      - input: ref('fact_learner_course_status')
-        format: sql
-        rows: |
-          select * from fact_learner_course_status
-      - input: ref('fact_learner_course_grade')
-        format: sql
-        rows: |
-          select * from fact_learner_course_grade
-      - input: ref('course_names')
-        format: sql
-        rows: |
-          select * from course_names
-      - input: ref('dim_user_pii')
-        format: sql
-        rows: |
-          select * from dim_user_pii
-    expect:
-      format: sql
-      rows: |
-        select * from fact_student_status_expected

models/video/schema.yml

@@ -250,9 +250,9 @@ models:
       - name: actor_id
         data_type: string
         description: "The xAPI actor identifier"
-      - name: video_id
-        data_type: String
-        description: "The xAPI object identifier"
+      - name: video_count
+        data_type: Int32
+        description: "Count of videos per course_key"
       - name: video_duration
         data_type: Int32
         description: "Total duration of the video"

models/video/watched_video_duration.sql

@@ -3,7 +3,7 @@
         materialized="materialized_view",
         schema=env_var("ASPECTS_XAPI_DATABASE", "xapi"),
         engine=get_engine("ReplacingMergeTree()"),
-        order_by="(org,course_key,actor_id,video_id)",
+        order_by="(org,course_key,actor_id)",
         post_hook="OPTIMIZE TABLE {{ this }} {{ on_cluster() }} FINAL",
     )
 }}
@@ -86,13 +86,19 @@ with
                 )
             )
             and b.start_emission_time > a.start_emission_time
+    ),
+    course_data as (
+        select org, course_key, count(distinct block_id) video_count
+        from {{ ref('dim_course_blocks') }}
+        where block_type = 'video'
+        group by org, course_key
     )
 select
-    org,
-    course_key,
+    course_data.org as org,
+    course_data.course_key as course_key,
     range.actor_id as actor_id,
-    video_id,
     video_duration,
+    video_count,
     sum(
         case
             when r1.actor_id = '' and r2.actor_id = ''
@@ -107,7 +113,8 @@ select
             else 0
         end
     ) as rewatched_time
-from range
+from course_data
+left join range on range.course_key = course_data.course_key
 left join rewatched r1 on range.event_id = r1.event_id1
 left join rewatched r2 on range.event_id = r2.event_id2
-group by org, course_key, actor_id, video_id, video_duration
+group by org, course_key, actor_id, video_count, video_duration