From f55f71c33571a0cde81589cd6cdfad0764af59d1 Mon Sep 17 00:00:00 2001
From: Carl Cervone <42869436+ccerv1@users.noreply.github.com>
Date: Wed, 19 Jun 2024 14:33:28 -0400
Subject: [PATCH] fix: include blockchain EOA artifacts in the artifacts table
 (#1679)

---
 .../intermediate/directory/int_artifacts.sql  | 44 ++++++++++++++++---
 1 file changed, 39 insertions(+), 5 deletions(-)

diff --git a/warehouse/dbt/models/intermediate/directory/int_artifacts.sql b/warehouse/dbt/models/intermediate/directory/int_artifacts.sql
index 623729c0d..a400d8f7c 100644
--- a/warehouse/dbt/models/intermediate/directory/int_artifacts.sql
+++ b/warehouse/dbt/models/intermediate/directory/int_artifacts.sql
@@ -1,9 +1,9 @@
 with all_artifacts as (
-  {# 
-    The `last_used` value is later used in this query to determine what the most
-    _current_ name is. However, oss-directory names are considered canonical so
-    `last_used` is only relevent for `git_user` artifacts.
+
+  {#
+    This grabs all the artifacts we know about from OSSD and from the contract discovery process.
   #}
+
   select
     artifact_source_id,
     artifact_source,
@@ -11,13 +11,47 @@ with all_artifacts as (
     artifact_url,
     artifact_name
   from {{ ref('int_artifacts_by_project') }}
+
   union all
+
+  {#
+    This grabs the universe of blockchain artifacts that have interacted with the contracts we care about from the events table.
+    TODO: this should be refactored when we "index the universe"
+  #}
+
+  select distinct
+    artifact_source_id,
+    artifact_source,
+    lower(artifact_source) as artifact_namespace,
+    artifact_source_id as artifact_name,
+    artifact_source_id as artifact_url
+  from (
+    select
+      from_artifact_source_id as artifact_source_id,
+      event_source as artifact_source
+    from {{ ref('int_events') }}
+    where event_type = 'CONTRACT_INVOCATION_DAILY_COUNT'
+    union all
+    select
+      to_artifact_source_id as artifact_source_id,
+      event_source as artifact_source
+    from {{ ref('int_events') }}
+    where event_type = 'CONTRACT_INVOCATION_DAILY_COUNT'
+  )
+
+  union all
+
+  {# 
+    This grabs the universe of GitHub users that have interacted with the repos we care about.
+    The `last_used` value is later used in this query to determine what the most _current_ name is. However, oss-directory names are considered canonical so `last_used` is only relevent for `git_user` artifacts.
+  #}
+
   select
     artifact_source_id,
     artifact_source,
     artifact_namespace,
     artifact_url,
-    MAX_BY(artifact_name, last_used) as artifact_name
+    max_by(artifact_name, last_used) as artifact_name
   from {{ ref('int_artifacts_history') }}
   group by
     artifact_source_id,