From 6c407cefdadaede139569836c6a93df6d168fe8c Mon Sep 17 00:00:00 2001
From: Carl Cervone <42869436+ccerv1@users.noreply.github.com>
Date: Fri, 19 Apr 2024 22:05:17 -0400
Subject: [PATCH] dbt: refactor projects mart to use int model (#1262)

* refactor: create int_projects

* update projects mart

* change comments
---
 .../directory/int_project_owners.sql          |  5 +--
 .../intermediate/directory/int_projects.sql   | 36 +++++++++++++++++++
 .../dbt/models/marts/directory/projects.sql   | 17 +++++----
 3 files changed, 47 insertions(+), 11 deletions(-)
 create mode 100644 warehouse/dbt/models/intermediate/directory/int_projects.sql

diff --git a/warehouse/dbt/models/intermediate/directory/int_project_owners.sql b/warehouse/dbt/models/intermediate/directory/int_project_owners.sql
index 0da2d8609..853ba66f5 100644
--- a/warehouse/dbt/models/intermediate/directory/int_project_owners.sql
+++ b/warehouse/dbt/models/intermediate/directory/int_project_owners.sql
@@ -6,13 +6,14 @@ WITH ranked_repos AS (
     ROW_NUMBER() OVER (
       PARTITION BY project_id ORDER BY star_count DESC
     ) AS row_number,
-    COUNT(DISTINCT owner) OVER (PARTITION BY project_id) AS num_github_owners
+    COUNT(DISTINCT owner) OVER (PARTITION BY project_id) AS count_github_owners
   FROM {{ ref('stg_ossd__repositories_by_project') }}
 )
 
 SELECT
   project_id,
-  num_github_owners,
+  count_github_owners,
   LOWER(owner) AS primary_github_owner
+  --TODO: is_git_organization
 FROM ranked_repos
 WHERE row_number = 1
diff --git a/warehouse/dbt/models/intermediate/directory/int_projects.sql b/warehouse/dbt/models/intermediate/directory/int_projects.sql
new file mode 100644
index 000000000..91b6d228f
--- /dev/null
+++ b/warehouse/dbt/models/intermediate/directory/int_projects.sql
@@ -0,0 +1,36 @@
+WITH ranked_repos AS (
+  SELECT
+    project_id,
+    owner,
+    star_count,
+    ROW_NUMBER() OVER (
+      PARTITION BY project_id ORDER BY star_count DESC
+    ) AS row_number,
+    COUNT(DISTINCT owner) OVER (PARTITION BY project_id) AS count_github_owners
+  FROM {{ ref('stg_ossd__repositories_by_project') }}
+),
+
+project_owners AS (
+  SELECT
+    project_id,
+    count_github_owners,
+    LOWER(owner) AS primary_github_owner
+  {# TODO: is_git_organization #}
+  FROM ranked_repos
+  WHERE row_number = 1
+)
+
+SELECT
+  p.id AS project_id,
+  p.slug AS project_slug,
+  {# TODO: description AS project_description #}
+  p.name AS project_name,
+  p.namespace AS namespace,
+  po.primary_github_owner,
+  po.count_github_owners,
+  ARRAY_LENGTH(JSON_EXTRACT_ARRAY(p.github)) AS count_github_artifacts,
+  ARRAY_LENGTH(JSON_EXTRACT_ARRAY(p.blockchain)) AS count_blockchain_artifacts,
+  ARRAY_LENGTH(JSON_EXTRACT_ARRAY(p.npm)) AS count_npm_artifacts
+FROM {{ ref('stg_ossd__current_projects') }} AS p
+LEFT JOIN project_owners AS po
+  ON p.id = po.project_id
diff --git a/warehouse/dbt/models/marts/directory/projects.sql b/warehouse/dbt/models/marts/directory/projects.sql
index 35526d240..68a388c00 100644
--- a/warehouse/dbt/models/marts/directory/projects.sql
+++ b/warehouse/dbt/models/marts/directory/projects.sql
@@ -5,12 +5,11 @@
 }}
 
 SELECT
-  id AS project_id,
-  slug AS project_slug,
-  -- description AS project_description,
-  name AS project_name,
-  namespace AS user_namespace,
-  ARRAY_LENGTH(JSON_EXTRACT_ARRAY(github)) AS count_github_artifacts,
-  ARRAY_LENGTH(JSON_EXTRACT_ARRAY(blockchain)) AS count_blockchain_artifacts,
-  ARRAY_LENGTH(JSON_EXTRACT_ARRAY(npm)) AS count_npm_artifacts
-FROM {{ ref('stg_ossd__current_projects') }}
+  project_id,
+  project_slug,
+  project_name,
+  count_github_owners,
+  count_github_artifacts,
+  count_blockchain_artifacts,
+  count_npm_artifacts
+FROM {{ ref('int_projects') }}