Merge pull request #5 from umccr/add-prj-to-cnt-link

Added raw schema link between Project hub to Contact hub

dev/src/load.sh

@@ -15,4 +15,5 @@ PGPASSWORD=dev psql -h 0.0.0.0 -d orcavault -U dev <<EOF
 \copy ods.metadata_manager_project from '/data/orcavault_ods_metadata_manager_project.csv' with (format csv, header true, delimiter ',');
 \copy ods.metadata_manager_libraryprojectlink from '/data/orcavault_ods_metadata_manager_libraryprojectlink.csv' with (format csv, header true, delimiter ',');
 \copy ods.metadata_manager_contact from '/data/orcavault_ods_metadata_manager_contact.csv' with (format csv, header true, delimiter ',');
+\copy ods.metadata_manager_projectcontactlink from '/data/orcavault_ods_metadata_manager_projectcontactlink.csv' with (format csv, header true, delimiter ',');
 EOF

dev/src/ods.sql

@@ -197,3 +197,10 @@ CREATE TABLE IF NOT EXISTS orcavault.ods.metadata_manager_contact
     description varchar,
     email       varchar(254)
 );
+
+CREATE TABLE IF NOT EXISTS orcavault.ods.metadata_manager_projectcontactlink
+(
+    id                 bigint,
+    contact_orcabus_id varchar not null,
+    project_orcabus_id varchar not null
+);

orcavault/models/raw/link_project_contact.sql

@@ -0,0 +1,50 @@
+with source as (
+
+    select project_name as project_id, project_owner as contact_id from {{ source('ods', 'data_portal_labmetadata') }}
+    union
+    select project_name as project_id, project_owner as contact_id from {{ source('ods', 'data_portal_limsrow') }}
+    union
+    select project_id, contact_id from {{ source('ods', 'metadata_manager_project') }} as prj
+        join {{ source('ods', 'metadata_manager_projectcontactlink') }} as lnk on lnk.project_orcabus_id = prj.orcabus_id
+        join {{ source('ods', 'metadata_manager_contact') }} as cnt on lnk.contact_orcabus_id = cnt.orcabus_id
+
+),
+
+cleaned as (
+
+    select
+        distinct trim(project_id) as project_id, trim(contact_id) as contact_id
+    from
+        source
+    where
+        (project_id is not null and project_id <> '') and
+        (contact_id is not null and contact_id <> '')
+
+),
+
+transformed as (
+
+    select
+        encode(sha256(cast(contact_id as bytea)), 'hex') as contact_hk,
+        encode(sha256(cast(project_id as bytea)), 'hex') as project_hk,
+        cast('{{ run_started_at }}' as timestamptz) as load_datetime,
+        (select 'lab') as record_source
+    from
+        cleaned
+
+),
+
+final as (
+
+    select
+        encode(sha256(concat(contact_hk, project_hk)::bytea), 'hex') as project_contact_hk,
+        contact_hk,
+        project_hk,
+        load_datetime,
+        record_source
+    from
+        transformed
+
+)
+
+select * from final

orcavault/models/raw/link_schema.yml

@@ -158,6 +158,32 @@ models:
       - name: record_source
         data_type: varchar(255)
 
+  - name: link_project_contact
+    config:
+      contract: { enforced: true }
+    constraints:
+      - type: primary_key
+        columns: [ project_contact_hk ]
+      - type: foreign_key
+        columns: [ contact_hk ]
+        to: ref('hub_contact')
+        to_columns: [ contact_hk ]
+      - type: foreign_key
+        columns: [ project_hk ]
+        to: ref('hub_project')
+        to_columns: [ project_hk ]
+    columns:
+      - name: project_contact_hk
+        data_type: char(64)
+      - name: contact_hk
+        data_type: char(64)
+      - name: project_hk
+        data_type: char(64)
+      - name: load_datetime
+        data_type: timestamptz
+      - name: record_source
+        data_type: varchar(255)
+
   - name: link_internal_to_external_subject
     config:
       contract: { enforced: true }