From 007f103ef45e081cc9e95295869da2391446c83f Mon Sep 17 00:00:00 2001
From: Jennifer Melot <jtmelot@gmail.com>
Date: Fri, 10 May 2024 15:34:19 -0400
Subject: [PATCH] Update organizations query to use new airtable data

Temporarily points to the parat_input_test dataset
---
 company_linkage/sql/organizations.sql | 98 ++++++++++++---------------
 1 file changed, 45 insertions(+), 53 deletions(-)

diff --git a/company_linkage/sql/organizations.sql b/company_linkage/sql/organizations.sql
index 5f1434d6..97f9eeb1 100644
--- a/company_linkage/sql/organizations.sql
+++ b/company_linkage/sql/organizations.sql
@@ -1,11 +1,27 @@
+WITH sp_500 AS (
+  SELECT DISTINCT
+    CSET_id
+  FROM
+    parat_input_test.groups
+  where
+    name = "S&P 500"
+),
+global_500 AS (
+  SELECT DISTINCT
+    CSET_id
+  FROM
+    parat_input_test.groups
+  where
+    name = "Global 500"
+)
 SELECT
   * REPLACE( (
     SELECT
-      ARRAY_AGG(STRUCT(alias_language,
+      ARRAY_AGG(STRUCT(language,
           alias))
     FROM (
       SELECT
-        DISTINCT alias_language,
+        DISTINCT language,
         alias
       FROM
         UNNEST(aliases) a )) AS aliases, (
@@ -32,92 +48,68 @@ SELECT
 FROM (
   SELECT
     CSET_id,
-    organizations_joined.name,
+    organizations.name,
     STRUCT(city,
-      province_state,
-      organizations_joined.country) AS location,
+      state_or_province,
+      organizations.country) AS location,
     website,
-    ARRAY_AGG(STRUCT(alias_language,
+    ARRAY_AGG(STRUCT(aliases.language,
         alias)) AS aliases,
     ARRAY_AGG(STRUCT(CASE
-          WHEN parent_acquisition IS TRUE THEN TRUE
+          WHEN parentage.is_acquisition_parent IS TRUE THEN TRUE
         ELSE
         FALSE
       END
         AS parent_acquisition,
         parent_name,
         parent_id)) AS parent,
-    ARRAY_AGG(DISTINCT permid IGNORE NULLS) AS permid,
-    ARRAY_AGG(STRUCT(exchange,
+    ARRAY_AGG(DISTINCT IF(source = "PermID", external_id, null) IGNORE NULLS) AS permid,
+    ARRAY_AGG(STRUCT(market as exchange,
         ticker)) AS market,
-    STRUCT(crunchbase_uuid,
-      crunchbase_url) AS crunchbase,
+    ARRAY_AGG(DISTINCT IF(source = "Crunchbase UUID", external_id, null) IGNORE NULLS) AS crunchbase_uuid,
+    ARRAY_AGG(DISTINCT IF(source = "Crunchbase URL", external_id, null) IGNORE NULLS) AS crunchbase_url,
     ARRAY_AGG(DISTINCT ror.id IGNORE NULLS) AS ror_id,
-    regex,
-    ARRAY_AGG(DISTINCT bgov_id IGNORE NULLS) AS BGOV_id,
-    linkedin,
-    CASE
-      WHEN in_sandp_500 IS TRUE THEN TRUE
-    ELSE
-    FALSE
-  END
-    AS in_sandp_500,
-    CASE
-      WHEN in_fortune_global_500 IS TRUE THEN TRUE
-    ELSE
-    FALSE
-  END
-    AS in_fortune_global_500,
-    ids_joined.comment
+    ARRAY_AGG(DISTINCT IF(source = "Regex", external_id, null) IGNORE NULLS) AS regex,
+    ARRAY_AGG(DISTINCT IF(source = "BGOV", external_id, null) IGNORE NULLS) AS BGOV_id,
+    ARRAY_AGG(DISTINCT IF(source = "LinkedIn", external_id, null) IGNORE NULLS) AS linkedin,
+    sp_500.CSET_id IS NOT NULL AS in_sandp_500,
+    global_500.CSET_id IS NOT NULL AS in_fortune_global_500
   FROM
-    parat_input.organizations_joined
-  LEFT JOIN
-    parat_input.alias_joined
-  USING
-    (CSET_id)
+    parat_input_test.organizations
   LEFT JOIN
-    parat_input.parent_joined
+    parat_input_test.aliases
   USING
     (CSET_id)
   LEFT JOIN
-    parat_input.permid_joined
+    parat_input_test.parentage
   USING
     (CSET_id)
   LEFT JOIN
-    parat_input.market_joined
+    parat_input_test.ids
   USING
     (CSET_id)
   LEFT JOIN
-    parat_input.ids_joined
+    parat_input_test.tickers
   USING
     (CSET_id)
   LEFT JOIN
-    parat_input.grid_joined
-  USING
-    (CSET_id)
+    gcp_cset_ror.ror
+  ON
+    TRIM(ids.external_id) = external_ids.GRID.all and ids.source = "GRID"
   LEFT JOIN
-    parat_input.bgov_validate
+    sp_500
   USING
     (CSET_id)
   LEFT JOIN
-    parat_input.linkedin_joined
+    global_500
   USING
     (CSET_id)
-  LEFT JOIN
-    gcp_cset_ror.ror
-  ON
-    TRIM(grid_joined.grid) = external_ids.GRID.all
   GROUP BY
     CSET_id,
     name,
     city,
-    province_state,
-    organizations_joined.country,
+    state_or_province,
+    organizations.country,
     website,
-    crunchbase_uuid,
-    crunchbase_url,
-    regex,
-    linkedin,
     in_sandp_500,
-    in_fortune_global_500,
-    comment)
\ No newline at end of file
+    in_fortune_global_500)
\ No newline at end of file