From 4de8b9e0f14a802735c28466866a752ba74ca9b5 Mon Sep 17 00:00:00 2001
From: Carl Cervone <42869436+ccerv1@users.noreply.github.com>
Date: Wed, 19 Jun 2024 12:48:45 -0400
Subject: [PATCH] dbt: identify contracts deployed by factories (#1675)

* dbt: identify contracts deployed by factories

* refactor int_contracts_by_project

* materialize int_contracts_by_project as table

---------

Co-authored-by: Reuven V. Gonzales <reuven.v.gonzales@gmail.com>
---
 .../directory/int_all_artifacts.sql           | 39 ++++++--
 .../directory/int_contracts_by_project.sql    | 88 ++++++++++++++-----
 2 files changed, 96 insertions(+), 31 deletions(-)

diff --git a/warehouse/dbt/models/intermediate/directory/int_all_artifacts.sql b/warehouse/dbt/models/intermediate/directory/int_all_artifacts.sql
index 4b47f191d..564726ecd 100644
--- a/warehouse/dbt/models/intermediate/directory/int_all_artifacts.sql
+++ b/warehouse/dbt/models/intermediate/directory/int_all_artifacts.sql
@@ -19,9 +19,7 @@ with ossd_artifacts as (
     artifact_name,
     artifact_url
   from {{ ref("int_artifacts_in_ossd_by_project") }}
-  where
-    artifact_type != 'DEPLOYER'
-    and artifact_source != 'ANY_EVM'
+  where artifact_type not in ('DEPLOYER', 'CONTRACT')
 ),
 
 verified_deployers as (
@@ -50,7 +48,7 @@ verified_contracts as (
   from {{ ref("int_contracts_by_project") }}
 ),
 
-all_normalized_artifacts as (
+onchain_artifacts as (
   select
     project_id,
     artifact_id,
@@ -60,7 +58,7 @@ all_normalized_artifacts as (
     artifact_namespace,
     artifact_name,
     artifact_url
-  from ossd_artifacts
+  from verified_deployers
   union all
   select
     project_id,
@@ -71,7 +69,34 @@ all_normalized_artifacts as (
     artifact_namespace,
     artifact_name,
     artifact_url
-  from verified_deployers
+  from verified_contracts
+),
+
+other_artifacts as (
+  select
+    project_id,
+    artifact_id,
+    artifact_source_id,
+    artifact_source,
+    artifact_type,
+    artifact_namespace,
+    artifact_name,
+    artifact_url
+  from ossd_artifacts
+  where artifact_id not in (select artifact_id from onchain_artifacts)
+),
+
+all_normalized_artifacts as (
+  select
+    project_id,
+    artifact_id,
+    artifact_source_id,
+    artifact_source,
+    artifact_type,
+    artifact_namespace,
+    artifact_name,
+    artifact_url
+  from other_artifacts
   union all
   select
     project_id,
@@ -82,7 +107,7 @@ all_normalized_artifacts as (
     artifact_namespace,
     artifact_name,
     artifact_url
-  from verified_contracts
+  from onchain_artifacts
 )
 
 select distinct
diff --git a/warehouse/dbt/models/intermediate/directory/int_contracts_by_project.sql b/warehouse/dbt/models/intermediate/directory/int_contracts_by_project.sql
index 0190e3808..2171a42c9 100644
--- a/warehouse/dbt/models/intermediate/directory/int_contracts_by_project.sql
+++ b/warehouse/dbt/models/intermediate/directory/int_contracts_by_project.sql
@@ -1,35 +1,75 @@
-with contracts as (
+{{
+  config(
+    materialized="table"
+  )
+}}
+with contracts_in_ossd as (
   select
-    block_timestamp,
-    transaction_hash,
-    network,
-    deployer_address,
-    contract_address
-  from {{ ref('int_derived_contracts') }}
+    project_id,
+    artifact_source as network,
+    artifact_name as contract_address
+  from {{ ref('int_artifacts_in_ossd_by_project') }}
+  where artifact_type = 'CONTRACT'
 ),
 
-deployers as (
+derived_contracts_by_project as (
   select
-    project_id,
-    artifact_name as deployer_address,
-    artifact_source as network
-  from {{ ref('int_deployers_by_project') }}
+    deployers_by_project.project_id,
+    derived_contracts.network,
+    derived_contracts.contract_address
+  from {{ ref('int_derived_contracts') }} as derived_contracts
+  left join {{ ref('int_deployers_by_project') }} as deployers_by_project
+    on
+      derived_contracts.deployer_address = deployers_by_project.artifact_name
+      and derived_contracts.network = deployers_by_project.artifact_source
+  where deployers_by_project.artifact_name is not null
+),
+
+unified_contracts as (
+  select distinct *
+  from (
+    select
+      project_id,
+      network,
+      contract_address
+    from contracts_in_ossd
+    union all
+    select
+      project_id,
+      network,
+      contract_address
+    from derived_contracts_by_project
+  )
 ),
 
 discovered_contracts as (
   select
-    deployers.project_id,
-    contracts.block_timestamp,
-    contracts.transaction_hash,
-    contracts.network,
-    contracts.deployer_address,
-    contracts.contract_address
-  from contracts
-  left join deployers
+    unified_contracts.project_id,
+    unified_contracts.network,
+    factories.contract_address
+  from {{ ref('int_factories') }} as factories
+  left join unified_contracts
     on
-      contracts.deployer_address = deployers.deployer_address
-      and contracts.network = deployers.network
-  where deployers.deployer_address is not null
+      factories.factory_address = unified_contracts.contract_address
+      and factories.network = unified_contracts.network
+  where unified_contracts.project_id is not null
+),
+
+contracts_by_project as (
+  select distinct *
+  from (
+    select
+      project_id,
+      network,
+      contract_address
+    from discovered_contracts
+    union all
+    select
+      project_id,
+      network,
+      contract_address
+    from unified_contracts
+  )
 )
 
 select
@@ -39,4 +79,4 @@ select
   contract_address as artifact_source_id,
   LOWER(network) as artifact_namespace,
   contract_address as artifact_name
-from discovered_contracts
+from contracts_by_project