Merge branch 'main' into mvtest

openedx · Nov 7, 2024 · 992d506 · 992d506
2 parents a57cd06 + 6e53552
commit 992d506
Show file tree

Hide file tree

Showing 8 changed files with 35 additions and 11 deletions.
diff --git a/models/enrollment/fact_enrollments.sql b/models/enrollment/fact_enrollments.sql
@@ -25,4 +25,6 @@ select
 from enrollments
 join {{ ref("course_names") }} courses on enrollments.course_key = courses.course_key
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (actor_id like 'mailto:%' and SUBSTRING(actor_id, 8) = users.email)
+    or actor_id = toString(users.external_user_id)
diff --git a/models/grading/fact_student_status.sql b/models/grading/fact_student_status.sql
@@ -29,4 +29,6 @@ join
     on fes.org = courses.org
     and fes.course_key = courses.course_key
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (actor_id like 'mailto:%' and SUBSTRING(actor_id, 8) = users.email)
+    or actor_id = toString(users.external_user_id)
diff --git a/models/navigation/fact_navigation.sql b/models/navigation/fact_navigation.sql
@@ -23,4 +23,6 @@ join
         and navigation.block_id = blocks.block_id
     )
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (actor_id like 'mailto:%' and SUBSTRING(actor_id, 8) = users.email)
+    or actor_id = toString(users.external_user_id)
diff --git a/models/navigation/fact_pageview_engagement.sql b/models/navigation/fact_pageview_engagement.sql
@@ -47,4 +47,6 @@ join
         and pv.block_id = course_blocks.block_id
     )
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(pv.actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (pv.actor_id like 'mailto:%' and SUBSTRING(pv.actor_id, 8) = users.email)
+    or pv.actor_id = toString(users.external_user_id)
diff --git a/models/problems/fact_problem_responses.sql b/models/problems/fact_problem_responses.sql
@@ -44,7 +44,9 @@ join
         and responses.problem_id = blocks.block_id
     )
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (actor_id like 'mailto:%' and SUBSTRING(actor_id, 8) = users.email)
+    or actor_id = toString(users.external_user_id)
 group by
     -- multi-part questions include an extra record for the response to the first
     -- part of the question. this group by clause eliminates the duplicate record

diff --git a/models/users/user_pii.sql b/models/users/user_pii.sql
@@ -4,7 +4,7 @@
         schema=env_var("ASPECTS_EVENT_SINK_DATABASE", "event_sink"),
         fields=[
             ("user_id", "Int32"),
-            ("external_user_id", "UUID"),
+            ("external_user_id", "String"),
             ("username", "String"),
             ("name", "String"),
             ("email", "String"),
@@ -25,9 +25,19 @@ with
         from {{ source("event_sink", "user_profile") }}
         group by user_id
     )
-select ex.user_id as user_id, ex.external_user_id, ex.username, up.name, up.email
-from {{ source("event_sink", "external_id") }} ex
-left outer join most_recent_user_profile mrup on mrup.user_id = ex.user_id
+select
+    ex.user_id as user_id,
+    if(
+        empty(ex.external_user_id),
+        concat('mailto:', email),
+        ex.external_user_id::String
+    ) as external_user_id,
+    up.username as username,
+    up.name as name,
+    up.email as email
+from most_recent_user_profile mrup
+left outer join
+    {{ source("event_sink", "external_id") }} ex on mrup.user_id = ex.user_id
 left outer join
     {{ source("event_sink", "user_profile") }} up
     on up.user_id = mrup.user_id

diff --git a/models/video/fact_video_engagement.sql b/models/video/fact_video_engagement.sql
@@ -46,4 +46,6 @@ join
         and ve.block_id = course_blocks.block_id
     )
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(ve.actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (ve.actor_id like 'mailto:%' and SUBSTRING(ve.actor_id, 8) = users.email)
+    or ve.actor_id = toString(users.external_user_id)
diff --git a/models/video/fact_video_plays.sql b/models/video/fact_video_plays.sql
@@ -42,4 +42,6 @@ join
     {{ ref("dim_course_blocks_extended") }} blocks
     on (plays.course_key = blocks.course_key and plays.video_id = blocks.block_id)
 left outer join
-    {{ ref("dim_user_pii") }} users on toUUID(actor_id) = users.external_user_id
+    {{ ref("dim_user_pii") }} users
+    on (actor_id like 'mailto:%' and SUBSTRING(actor_id, 8) = users.email)
+    or actor_id = toString(users.external_user_id)