Add teacher details directly to the users table

lms/data_tasks/report/create_from_scratch/02_entities/01_users/01_raw_users/01_create_view.jinja2.sql

@@ -0,0 +1,15 @@
+DROP MATERIALIZED VIEW IF EXISTS report.raw_users CASCADE;
+
+-- The building blocks of a user. We won't export this. See
+-- `*_entities_decorated/*_users` for the version for external consumers.
+CREATE MATERIALIZED VIEW report.raw_users AS (
+    SELECT
+         id,
+         username,
+         registered_date::date
+    FROM h.user AS users
+    WHERE
+        users.authority = '{{ region.authority }}'
+        -- users.authority = 'lms.hypothes.is'
+    ORDER BY registered_date
+) WITH NO DATA;

lms/data_tasks/report/create_from_scratch/02_entities/01_users/01_raw_users/02_initial_fill.sql

@@ -0,0 +1,8 @@
+DROP INDEX IF EXISTS report.raw_users_id_idx;
+
+REFRESH MATERIALIZED VIEW report.raw_users;
+
+ANALYSE report.raw_users;
+
+-- A unique index is mandatory for concurrent updates used in the refresh
+CREATE UNIQUE INDEX raw_users_id_idx ON report.raw_users (id);

lms/data_tasks/report/create_from_scratch/02_entities/01_users/02_user_map/01_create_view.sql

@@ -4,11 +4,11 @@ DROP MATERIALIZED VIEW IF EXISTS report.user_map CASCADE;
 CREATE MATERIALIZED VIEW report.user_map AS (
     SELECT
         lms_users.id AS lms_user_id,
-        users.id AS user_id,
+        raw_users.id AS user_id,
         application_instances.organization_id
     FROM "user" AS lms_users
-    JOIN report.users ON
-        users.username = SUBSTRING(SPLIT_PART(lms_users.h_userid, '@', 1), 6)
+    JOIN report.raw_users ON
+        raw_users.username = SUBSTRING(SPLIT_PART(lms_users.h_userid, '@', 1), 6)
     JOIN application_instances ON
         lms_users.application_instance_id = application_instances.id
     ORDER BY lms_user_id, user_id

lms/data_tasks/report/create_from_scratch/05_entities_decorated/01_users/01_create_view.jinja2.sql

@@ -0,0 +1,58 @@
+DROP MATERIALIZED VIEW IF EXISTS report.users CASCADE;
+
+CREATE MATERIALIZED VIEW report.users AS (
+    WITH
+        user_details AS (
+            SELECT DISTINCT
+                user_map.user_id,
+
+                -- The order here should ensure empty strings and nulls are
+                -- sorted to the back, so the "first" value should be non-null
+                -- if possible
+                FIRST_VALUE(lms_user.display_name) OVER (
+                    PARTITION BY user_map.user_id
+                    ORDER BY lms_user.display_name
+                    RANGE BETWEEN
+                        UNBOUNDED PRECEDING AND
+                        UNBOUNDED FOLLOWING
+                ) AS display_name,
+
+                FIRST_VALUE(lms_user.email) OVER (
+                    PARTITION BY user_map.user_id
+                    ORDER BY lms_user.email
+                    RANGE BETWEEN
+                        UNBOUNDED PRECEDING AND
+                        UNBOUNDED FOLLOWING
+                ) AS email,
+
+                true AS is_teacher
+            FROM report.user_map
+            JOIN "user" AS lms_user ON
+                lms_user.id = user_map.lms_user_id
+            WHERE
+                -- Ensure we are only providing contact details for users who
+                -- are a teacher in at least one context
+                user_map.user_id IN (
+                    SELECT DISTINCT(user_id)
+                    FROM report.organization_roles
+                    WHERE role = 'teacher'
+                )
+        )
+
+    SELECT
+        raw_users.id,
+        user_details.display_name,
+        user_details.email,
+        raw_users.username,
+        CASE
+            WHEN user_details.is_teacher IS true THEN true ELSE false
+        END AS is_teacher,
+        -- Add a column which indicates that a user is a teacher in at least
+        -- one context. This helps us notice / demonstrate we are only keeping
+        -- teacher contact info. Keeping student contact info out of region is
+        -- not permitted.
+        raw_users.registered_date
+    FROM report.raw_users
+    LEFT OUTER JOIN user_details ON
+        user_details.user_id = raw_users.id
+) WITH NO DATA;

lms/data_tasks/report/create_from_scratch/02_entities/01_users/01_users/02_initial_fill.sql → lms/data_tasks/report/create_from_scratch/05_entities_decorated/01_users/02_initial_fill.sql

@@ -1,10 +1,8 @@
 DROP INDEX IF EXISTS report.users_id_idx;
-DROP INDEX IF EXISTS report.users_registered_date_idx;
 
 REFRESH MATERIALIZED VIEW report.users;
 
 ANALYSE report.users;
 
 -- A unique index is mandatory for concurrent updates used in the refresh
 CREATE UNIQUE INDEX users_id_idx ON report.users (id);
-CREATE INDEX users_registered_date_idx ON report.users USING BRIN (registered_date);

lms/data_tasks/report/create_from_scratch/05_entities_decorated/02_users_sensitive/01_create_view.sql

@@ -0,0 +1,8 @@
+DROP VIEW IF EXISTS report.users_sensitive CASCADE;
+
+-- Create a plain view to act as a shim while we replace `users_sensitive`
+CREATE VIEW report.users_sensitive AS (
+    SELECT
+        id, email, display_name
+    FROM report.users
+);

lms/data_tasks/report/refresh/01_entities_refresh.sql

@@ -1,5 +1,5 @@
-REFRESH MATERIALIZED VIEW CONCURRENTLY report.users;
-ANALYSE report.users;
+REFRESH MATERIALIZED VIEW CONCURRENTLY report.raw_users;
+ANALYSE report.raw_users;
 
 REFRESH MATERIALIZED VIEW CONCURRENTLY report.user_map;
 ANALYSE report.user_map;

lms/data_tasks/report/refresh/04_entities_decorated.sql

@@ -0,0 +1,2 @@
+REFRESH MATERIALIZED VIEW CONCURRENTLY report.users;
+ANALYSE report.users;