add language to repo stats (#1455)

warehouse/dbt/models/intermediate/metrics/int_repo_metrics_by_project.sql

@@ -5,6 +5,7 @@ with repo_artifact as (
     fork_count,
     star_count,
     license_spdx_id,
+    language,
     watcher_count,
     CAST(id as STRING) as artifact_source_id,
     LOWER(owner) as artifact_namespace,
@@ -18,6 +19,7 @@ repo_snapshot as (
     artifact_namespace,
     artifact_name,
     license_spdx_id,
+    language,
     is_fork,
     fork_count,
     star_count,
@@ -51,6 +53,7 @@ select
   repo_snapshot.fork_count,
   repo_snapshot.star_count,
   repo_snapshot.watcher_count,
+  repo_snapshot.language,
   repo_snapshot.license_spdx_id,
   repo_stats.first_commit_time,
   repo_stats.last_commit_time,

warehouse/dbt/models/marts/superchain/rf4_repo_stats_by_project.sql

@@ -11,6 +11,7 @@ select
   --int_repo_metrics_by_project.last_commit_time,
   --int_repo_metrics_by_project.days_with_commits_count,
   --int_repo_metrics_by_project.contributors_to_repo_count,
+  int_repo_metrics_by_project.language,
   int_repo_metrics_by_project.license_spdx_id,
   case
     {# TODO: Review licenses https://spdx.org/licenses/ for OSI Approved #}
@@ -22,9 +23,11 @@ select
       'ISC', '0BSD', 'NCSA', 'Zlib'
     ) then 'Permissive'
     when int_repo_metrics_by_project.license_spdx_id in (
-      'BSD-4-Clause', 'WTFPL', 'NOASSERTION',
+      'BSD-4-Clause', 'WTFPL',
       'CC0-1.0', 'CC-BY-SA-4.0', 'CC-BY-4.0'
     ) then 'Restrictive'
+    when int_repo_metrics_by_project.license_spdx_id = 'NOASSERTION'
+      then 'Custom'
     else 'Unspecified'
   end as license_type
 from {{ ref('int_repo_metrics_by_project') }}

warehouse/dbt/models/staging/oss-directory/stg_ossd__current_repositories.sql

@@ -20,6 +20,7 @@ select
   repositories.is_fork,
   repositories.license_name,
   repositories.license_spdx_id,
+  repositories.language,
   repositories._cq_sync_time as `sync_time`
 from {{ oso_source('ossd', 'repositories') }} as repositories
 where _cq_sync_time = (select * from most_recent_sync)