Merge pull request #7 from clingen-data-model/cvc-script-updates

clinvar cvc function and script updates in prep of move from stage to…

scripts/clinvar-curartion/cvc-baseline-annotations-as-of-func.sql

@@ -1,37 +1,29 @@
 CREATE OR REPLACE TABLE FUNCTION `clinvar_curator.cvc_baseline_annotations_as_of`(as_of_date DATE, incl_finalized BOOL, only_latest BOOL) AS (
-WITH anno AS 
-    (
-      SELECT
-        rel.release_date,
-        CAST(UNIX_MILLIS(annotation_date) AS STRING) as annotation_id,
-        a.vcv_id as vcv_axn,
-        SPLIT(a.scv_id,'.')[OFFSET(0)] AS scv_id,
-        CAST(SPLIT(a.scv_id,'.')[OFFSET(1)] AS INT64) AS scv_ver,
-        CAST(a.variation_id AS String) AS variation_id,
-        CAST(a.submitter_id AS String) AS submitter_id,
-        LOWER(a.action) AS action,
-        SPLIT(a.curator_email,'@')[OFFSET(0)] AS curator,
-        a.annotation_date AS annotated_on,
-        DATE(a.annotation_date) AS annotated_date,
-        TIME(a.annotation_date) AS annotated_time_utc,
-        a.reason,
-        a.notes,
-        SPLIT(a.vcv_id,'.')[OFFSET(0)] AS vcv_id,
-        CAST(SPLIT(a.vcv_id,'.')[OFFSET(1)] AS INT64) AS vcv_ver,
-        -- if there are no other scv_id annotations after when orderd by annotation date then it is the latest
-        (
-          COUNT(a.annotation_date) 
-          OVER (
-            PARTITION BY SPLIT(a.scv_id,'.')[OFFSET(0)] 
-            ORDER BY a.annotation_date 
-            ROWS BETWEEN 1 FOLLOWING AND UNBOUNDED FOLLOWING
-          ) = 0
-        ) AS is_latest
-
-      FROM `clinvar_curator.clinvar_annotations` a,
-        `clinvar_ingest.schema_on`(as_of_date) rel
-      WHERE 
-        DATE(a.annotation_date) <= as_of_date
+WITH 
+  anno AS (
+    SELECT
+      rel.release_date,
+      av.annotation_id,
+      av.vcv_axn,
+      av.scv_id,
+      av.scv_ver,
+      av.variation_id,
+      av.submitter_id,
+      av.action,
+      av.curator,
+      av.annotated_on,
+      av.annotated_date,
+      av.annotated_time_utc,
+      av.reason,
+      av.notes,
+      av.vcv_id,
+      av.vcv_ver,
+      av.is_latest,
+      av.annotation_label
+    FROM `clinvar_curator.cvc_annotations_view` av,
+      `clinvar_ingest.schema_on`(as_of_date) rel
+    WHERE 
+      av.annotated_date <= as_of_date
   ),
   reviewed_anno AS (
     SELECT
@@ -67,18 +59,6 @@ WITH anno AS
       rev.notes,
       sbm.batch_id
   ),
-  anno_label as (
-    SELECT 
-      a.annotation_id,
-      FORMAT(
-        '%t (%s) %s: %s',
-        a.annotated_date, 
-        IFNULL(a.curator,'n/a'), 
-        IF(a.action ='flagging candidate','flag',IF(a.action = 'no change', 'no chg', 'n/a or unk' )), 
-        LEFT(IFNULL(a.reason,'n/a'), 20)||IF(LENGTH(a.reason) > 20,'...','')
-      ) as annotation_label
-    FROM anno a
-  ),
   ra_priors AS (
     SELECT
       a.annotation_id,
@@ -89,7 +69,7 @@ WITH anno AS
         FORMAT(
           'v%i %s %s',
           prior_a.scv_ver,
-          prior_a_label.annotation_label, 
+          prior_a.annotation_label, 
           if(prior_ra.review_label is not null, FORMAT('[ %s ]',prior_ra.review_label), '')
         ),
         '\n' 
@@ -105,10 +85,6 @@ WITH anno AS
     ON 
       prior_a.scv_id = a.scv_id and 
       prior_a.annotation_id < a.annotation_id
-    JOIN anno_label prior_a_label
-    ON
-      prior_a.annotation_id = prior_a_label.annotation_id
-
     LEFT JOIN reviewed_anno as prior_ra
     ON
       prior_ra.annotation_id = prior_a.annotation_id
@@ -140,13 +116,7 @@ WITH anno AS
     a.reason,
     a.notes,
     a.submitter_id,
-    FORMAT(
-      '%t (%s) %s: %s',
-      a.annotated_date, 
-      IFNULL(a.curator,'n/a'), 
-      IF(a.action ='flagging candidate','flag',IF(a.action = 'no change', 'no chg', 'n/a or unk' )), 
-      LEFT(IFNULL(a.reason,'n/a'), 20)||IF(LENGTH(a.reason) > 20,'...','')
-    ) as annotation_label,
+    a.annotation_label,
     a.is_latest,
     -- finalized rev info
     ra.annotation_id as finalized_review_id,