Merge pull request #130 from nelc/and/fix_external_id_searches

fix: parse external_id to string instead of actor id to uuid and allo…
openedx · Oct 31, 2024 · 4139d69 · 4139d69
2 parents ee5b022 + b4dc921
commit 4139d69
Show file tree

Hide file tree

Showing 7 changed files with 21 additions and 7 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/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)