feat: additional rf4 repo eligibility checks (#1535)

* feat: additional rf4 repo eligibility checks * chore: add repo checks to overall project checks * fix: clean up todos * fix: handling empty arrays
opensource-observer · May 27, 2024 · e1f317c · e1f317c
1 parent ddbfbdb
commit e1f317c
Show file tree

Hide file tree

Showing 5 changed files with 115 additions and 56 deletions.
diff --git a/warehouse/dbt/models/intermediate/metrics/int_repo_metrics_by_project.sql b/warehouse/dbt/models/intermediate/metrics/int_repo_metrics_by_project.sql
@@ -34,7 +34,8 @@ repo_stats as (
     MIN(time) as first_commit_time,
     MAX(time) as last_commit_time,
     COUNT(distinct TIMESTAMP_TRUNC(time, day)) as days_with_commits_count,
-    COUNT(distinct from_artifact_id) as contributors_to_repo_count
+    COUNT(distinct from_artifact_id) as contributors_to_repo_count,
+    SUM(amount) as commit_count
   from {{ ref('int_events_to_project') }}
   where event_type = 'COMMIT_CODE'
   group by
@@ -58,7 +59,8 @@ select
   repo_stats.first_commit_time,
   repo_stats.last_commit_time,
   repo_stats.days_with_commits_count,
-  repo_stats.contributors_to_repo_count
+  repo_stats.contributors_to_repo_count,
+  repo_stats.commit_count
 from {{ ref('int_artifacts_by_project') }}
 left join repo_snapshot
   on int_artifacts_by_project.artifact_id = repo_snapshot.artifact_id

diff --git a/warehouse/dbt/models/marts/superchain/metrics/rf4_monthly_active_addresses.sql b/warehouse/dbt/models/marts/superchain/metrics/rf4_monthly_active_addresses.sql
@@ -1,3 +1,4 @@
+{# TODO: double check the math on total_months #}
 with txns as (
   select
     project_id,
@@ -21,9 +22,7 @@ maas as (
 ),
 
 total_months as (
-  select
-    {# TODO: double check this math #}
-    (DATE_DIFF(max_month, min_month, day) + 30) / 30 as months
+  select (DATE_DIFF(max_month, min_month, day) + 30) / 30 as months
   from (
     select
       MIN(bucket_month) as min_month,

diff --git a/warehouse/dbt/models/marts/superchain/metrics/rf4_trusted_monthly_active_users.sql b/warehouse/dbt/models/marts/superchain/metrics/rf4_trusted_monthly_active_users.sql
@@ -1,3 +1,4 @@
+{# TODO: double check the math on total_months #}
 with txns as (
   select
     project_id,
@@ -22,9 +23,7 @@ maus as (
 ),
 
 total_months as (
-  select
-    {# TODO: double check this math #}
-    (DATE_DIFF(max_month, min_month, day) + 30) / 30 as months
+  select (DATE_DIFF(max_month, min_month, day) + 30) / 30 as months
   from (
     select
       MIN(bucket_month) as min_month,

diff --git a/warehouse/dbt/models/marts/superchain/rf4_project_verification.sql b/warehouse/dbt/models/marts/superchain/rf4_project_verification.sql
@@ -5,10 +5,23 @@
   - Review thresholds for unique_addresses, date_first_transaction, and
     days_with_onchain_activity_in_range.
   - Filter on contracts that are linked to Retro Funding applications (from Agora data)
-  - Integrate with repo_stats_by_project to check licensing and other repo requirements.
 #}
 
-with project_stats as (
+with repo_stats as (
+  select
+    project_id,
+    ARRAY_AGG(repo_name) as eligible_repos
+  from (
+    select
+      project_id,
+      CONCAT(artifact_namespace, '/', artifact_name) as repo_name
+    from {{ ref('rf4_repo_stats_by_project') }}
+    where approval_status = 'approved'
+  )
+  group by project_id
+),
+
+onchain_stats as (
   select
     project_id,
     project_name,
@@ -23,23 +36,38 @@ with project_stats as (
   from {{ ref('rf4_events_daily_to_project') }}
   where
     event_type = 'CONTRACT_INVOCATION_SUCCESS_DAILY_COUNT'
-    and bucket_day >= '2024-01-01'
+    and bucket_day >= '2023-10-01'
   group by
     project_id,
     project_name
 ),
 
-tests as (
+project_stats as (
+  select
+    onchain_stats.*,
+    COALESCE(repo_stats.eligible_repos, array<STRING>[]) as eligible_repos
+  from onchain_stats
+  left join repo_stats
+    on onchain_stats.project_id = repo_stats.project_id
+  left join {{ ref('projects_by_collection_v1') }}
+    on onchain_stats.project_id = projects_by_collection_v1.project_id
+  where
+    projects_by_collection_v1.collection_name = 'op-onchain'
+),
+
+checks as (
   select
     project_id,
     project_name,
+    eligible_repos,
     unique_addresses,
     date_first_transaction,
     days_with_onchain_activity_in_range,
-    unique_addresses >= 420 as test_unique_addresses,
-    date_first_transaction < '2024-03-01' as test_date_first_transaction,
+    ARRAY_LENGTH(eligible_repos) >= 1 as check_eligible_repos,
+    unique_addresses >= 420 as check_unique_addresses,
+    date_first_transaction < '2024-03-01' as check_date_first_transaction,
     days_with_onchain_activity_in_range >= 10
-      as test_days_with_onchain_activity_in_range
+      as check_days_with_onchain_activity_in_range
   from project_stats
 )
 
@@ -49,9 +77,14 @@ select
   unique_addresses,
   date_first_transaction,
   days_with_onchain_activity_in_range,
-  test_unique_addresses,
-  test_date_first_transaction,
-  test_days_with_onchain_activity_in_range,
-  test_unique_addresses and test_date_first_transaction
-  and test_days_with_onchain_activity_in_range as test_all
-from tests
+  check_eligible_repos,
+  check_unique_addresses,
+  check_date_first_transaction,
+  check_days_with_onchain_activity_in_range,
+  (
+    check_eligible_repos
+    and check_unique_addresses
+    and check_date_first_transaction
+    and check_days_with_onchain_activity_in_range
+  ) as meets_all_requirements
+from checks
diff --git a/warehouse/dbt/models/marts/superchain/rf4_repo_stats_by_project.sql b/warehouse/dbt/models/marts/superchain/rf4_repo_stats_by_project.sql
@@ -1,42 +1,68 @@
+{# TODO: Review licenses https://spdx.org/licenses/ for OSI Approved #}
+{# TODO: update with actual collection for RF4 #}
+with repo_snapshot as (
+  select
+    project_id,
+    artifact_namespace,
+    artifact_name,
+    is_fork,
+    fork_count,
+    star_count,
+    first_commit_time,
+    last_commit_time,
+    days_with_commits_count,
+    commit_count,
+    language,
+    license_spdx_id,
+    case
+      when license_spdx_id in (
+        'MIT', 'MIT-0', 'Apache-2.0', 'Unlicense',
+        'BSD-2-Clause', 'BSD-3-Clause', 'BSD-3-Clause-Clear',
+        'AGPL-3.0', 'GPL-3.0', 'LGPL-3.0', 'GPL-2.0', 'MPL-2.0', 'LGPL-2.1',
+        'OFL-1.1', 'EPL-1.0', 'EPL-2.0', 'OFL-1.1', 'EUPL-1.2', 'OSL-3.0',
+        'ISC', '0BSD', 'NCSA', 'Zlib'
+      ) then 'Permissive'
+      when license_spdx_id in (
+        'BSD-4-Clause', 'WTFPL',
+        'CC0-1.0', 'CC-BY-SA-4.0', 'CC-BY-4.0'
+      ) then 'Restrictive'
+      when license_spdx_id = 'NOASSERTION'
+        then 'Custom'
+      else 'Unspecified'
+    end as license_type
+  from {{ ref('int_repo_metrics_by_project') }}
+)
+
 select
-  int_repo_metrics_by_project.project_id,
-  int_repo_metrics_by_project.artifact_id,
+  repo_snapshot.project_id,
   projects_v1.project_name,
-  int_repo_metrics_by_project.artifact_namespace,
-  int_repo_metrics_by_project.artifact_name,
-  int_repo_metrics_by_project.is_fork,
-  int_repo_metrics_by_project.fork_count,
-  int_repo_metrics_by_project.star_count,
-  --int_repo_metrics_by_project.first_commit_time,
-  --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,
+  repo_snapshot.artifact_namespace,
+  repo_snapshot.artifact_name,
+  repo_snapshot.is_fork,
+  repo_snapshot.fork_count,
+  repo_snapshot.star_count,
+  repo_snapshot.first_commit_time,
+  repo_snapshot.last_commit_time,
+  repo_snapshot.days_with_commits_count,
+  repo_snapshot.commit_count,
+  repo_snapshot.language,
+  repo_snapshot.license_spdx_id,
+  repo_snapshot.license_type,
   case
-    {# TODO: Review licenses https://spdx.org/licenses/ for OSI Approved #}
-    when int_repo_metrics_by_project.license_spdx_id in (
-      'MIT', 'MIT-0', 'Apache-2.0', 'Unlicense',
-      'BSD-2-Clause', 'BSD-3-Clause', 'BSD-3-Clause-Clear',
-      'AGPL-3.0', 'GPL-3.0', 'LGPL-3.0', 'GPL-2.0', 'MPL-2.0', 'LGPL-2.1',
-      'OFL-1.1', 'EPL-1.0', 'EPL-2.0', 'OFL-1.1', 'EUPL-1.2', 'OSL-3.0',
-      'ISC', '0BSD', 'NCSA', 'Zlib'
-    ) then 'Permissive'
-    when int_repo_metrics_by_project.license_spdx_id in (
-      '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') }}
+    when (
+      repo_snapshot.commit_count >= 10
+      and repo_snapshot.days_with_commits_count >= 3
+      and repo_snapshot.first_commit_time < '2024-05-01'
+      and repo_snapshot.star_count >= 10
+      and repo_snapshot.language in ('Solidity', 'JavaScript', 'TypeScript')
+    ) then 'approved'
+    else 'review'
+  end as approval_status
+from repo_snapshot
 left join {{ ref('projects_v1') }}
-  on int_repo_metrics_by_project.project_id = projects_v1.project_id
+  on repo_snapshot.project_id = projects_v1.project_id
 left join {{ ref('projects_by_collection_v1') }}
-  on
-    int_repo_metrics_by_project.project_id
-    = projects_by_collection_v1.project_id
+  on repo_snapshot.project_id = projects_by_collection_v1.project_id
 where
-  {# TODO: update with actual collection for RF4 #}
   projects_by_collection_v1.collection_name = 'op-onchain'
+  and repo_snapshot.license_type != 'Unspecified'