Merge pull request #7 from umccr/add-int-to-ext-smp-link

Added raw schema link between Sample hub to ExternalSample hub

orcavault/models/raw/link_internal_to_external_sample.sql

@@ -0,0 +1,48 @@
+with source as (
+
+    select sample_id, external_sample_id from {{ source('ods', 'data_portal_labmetadata') }}
+    union
+    select sample_id, external_sample_id from {{ source('ods', 'data_portal_limsrow') }}
+    union
+    select sample_id, external_sample_id from {{ source('ods', 'metadata_manager_sample') }}
+
+),
+
+cleaned as (
+
+    select
+        distinct sample_id, trim(external_sample_id) as external_sample_id
+    from
+        source
+    where
+        (sample_id is not null and sample_id <> '') and
+        (external_sample_id is not null and external_sample_id <> '')
+
+),
+
+transformed as (
+
+    select
+        encode(sha256(cast(external_sample_id as bytea)), 'hex') as external_sample_hk,
+        encode(sha256(cast(sample_id as bytea)), 'hex') as sample_hk,
+        cast('{{ run_started_at }}' as timestamptz) as load_datetime,
+        (select 'lab') as record_source
+    from
+        cleaned
+
+),
+
+final as (
+
+    select
+        encode(sha256(concat(external_sample_hk, sample_hk)::bytea), 'hex') as internal_external_sample_hk,
+        external_sample_hk,
+        sample_hk,
+        load_datetime,
+        record_source
+    from
+        transformed
+
+)
+
+select * from final

orcavault/models/raw/link_schema.yml

@@ -235,3 +235,29 @@ models:
         data_type: timestamptz
       - name: record_source
         data_type: varchar(255)
+
+  - name: link_internal_to_external_sample
+    config:
+      contract: { enforced: true }
+    constraints:
+      - type: primary_key
+        columns: [ internal_external_sample_hk ]
+      - type: foreign_key
+        columns: [ external_sample_hk ]
+        to: ref('hub_external_sample')
+        to_columns: [ external_sample_hk ]
+      - type: foreign_key
+        columns: [ sample_hk ]
+        to: ref('hub_sample')
+        to_columns: [ sample_hk ]
+    columns:
+      - name: internal_external_sample_hk
+        data_type: char(64)
+      - name: external_sample_hk
+        data_type: char(64)
+      - name: sample_hk
+        data_type: char(64)
+      - name: load_datetime
+        data_type: timestamptz
+      - name: record_source
+        data_type: varchar(255)