From 52147fe778aaef71ffd57b6d75e1aae5b4ddba3d Mon Sep 17 00:00:00 2001
From: Reuven Gonzales <reuven@karibalabs.co>
Date: Sat, 23 Nov 2024 01:56:50 +0900
Subject: [PATCH] Auxiliary events table for issues (#2512)

* Auxiliary events table for issues

* Create mart for auxiliary table
---
 .../events/int_events_aux_issues.sql          | 123 ++++++++++++++++++
 ...eries_events_aux_issues_by_artifact_v0.sql |  16 +++
 2 files changed, 139 insertions(+)
 create mode 100644 warehouse/dbt/models/intermediate/events/int_events_aux_issues.sql
 create mode 100644 warehouse/dbt/models/marts/events/timeseries_events_aux_issues_by_artifact_v0.sql

diff --git a/warehouse/dbt/models/intermediate/events/int_events_aux_issues.sql b/warehouse/dbt/models/intermediate/events/int_events_aux_issues.sql
new file mode 100644
index 000000000..b3e42817f
--- /dev/null
+++ b/warehouse/dbt/models/intermediate/events/int_events_aux_issues.sql
@@ -0,0 +1,123 @@
+{{
+  config(
+    materialized='table',
+    partition_by={
+      "field": "time",
+      "data_type": "timestamp",
+      "granularity": "day",
+    },
+    meta={
+      'sync_to_db': False
+    }
+  )
+}}
+
+with github_comments as (
+  select -- noqa: ST06
+    created_at as `time`,
+    type as event_type,
+    CAST(id as STRING) as event_source_id,
+    "GITHUB" as event_source,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(1)]
+      as to_name,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(0)]
+      as to_namespace,
+    "REPOSITORY" as to_type,
+    CAST(repository_id as STRING) as to_artifact_source_id,
+    actor_login as from_name,
+    actor_login as from_namespace,
+    "GIT_USER" as from_type,
+    CAST(actor_id as STRING) as from_artifact_source_id,
+    `number` as issue_number
+  from {{ ref('stg_github__comments') }}
+),
+
+github_issues as (
+  select -- noqa: ST06
+    created_at as `time`,
+    type as event_type,
+    CAST(id as STRING) as event_source_id,
+    "GITHUB" as event_source,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(1)]
+      as to_name,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(0)]
+      as to_namespace,
+    "REPOSITORY" as to_type,
+    CAST(repository_id as STRING) as to_artifact_source_id,
+    actor_login as from_name,
+    actor_login as from_namespace,
+    "GIT_USER" as from_type,
+    CAST(actor_id as STRING) as from_artifact_source_id,
+    `number` as issue_number
+  from {{ ref('stg_github__issues') }}
+),
+
+github_pull_requests as (
+  select -- noqa: ST06
+    created_at as `time`,
+    type as event_type,
+    CAST(id as STRING) as event_source_id,
+    "GITHUB" as event_source,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(1)]
+      as to_name,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(0)]
+      as to_namespace,
+    "REPOSITORY" as to_type,
+    CAST(repository_id as STRING) as to_artifact_source_id,
+    actor_login as from_name,
+    actor_login as from_namespace,
+    "GIT_USER" as from_type,
+    CAST(actor_id as STRING) as from_artifact_source_id,
+    `number` as issue_number
+  from {{ ref('stg_github__pull_requests') }}
+),
+
+github_pull_request_merge_events as (
+  select -- noqa: ST06
+    created_at as `time`,
+    type as event_type,
+    CAST(id as STRING) as event_source_id,
+    "GITHUB" as event_source,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(1)]
+      as to_name,
+    SPLIT(REPLACE(repository_name, "@", ""), "/")[SAFE_OFFSET(0)]
+      as to_namespace,
+    "REPOSITORY" as to_type,
+    CAST(repository_id as STRING) as to_artifact_source_id,
+    actor_login as from_name,
+    actor_login as from_namespace,
+    "GIT_USER" as from_type,
+    CAST(actor_id as STRING) as from_artifact_source_id,
+    `number` as issue_number
+  from {{ ref('stg_github__pull_request_merge_events') }}
+),
+
+issue_events as (
+  select
+    time,
+    event_type,
+    event_source_id,
+    event_source,
+    {{ oso_id("event_source", "to_artifact_source_id") }} as to_artifact_id,
+    {{ oso_id("event_source", "from_artifact_source_id") }} as from_artifact_id,
+    issue_number
+  from (
+    select * from github_issues
+    union all
+    select * from github_pull_requests
+    union all
+    select * from github_pull_request_merge_events
+    union all
+    select * from github_comments
+  )
+)
+
+select
+  time,
+  to_artifact_id,
+  from_artifact_id,
+  issue_number,
+  UPPER(event_type) as event_type,
+  CAST(event_source_id as STRING) as event_source_id,
+  UPPER(event_source) as event_source
+from issue_events
diff --git a/warehouse/dbt/models/marts/events/timeseries_events_aux_issues_by_artifact_v0.sql b/warehouse/dbt/models/marts/events/timeseries_events_aux_issues_by_artifact_v0.sql
new file mode 100644
index 000000000..b8a708864
--- /dev/null
+++ b/warehouse/dbt/models/marts/events/timeseries_events_aux_issues_by_artifact_v0.sql
@@ -0,0 +1,16 @@
+{{ 
+  config(meta = {
+    'sync_to_db': True,
+    'order_by': [ 'event_source', 'event_type', 'to_artifact_id', 'time' ]
+  }) 
+}}
+
+select
+  time,
+  to_artifact_id,
+  from_artifact_id,
+  event_type,
+  event_source_id,
+  event_source,
+  issue_number
+from {{ ref('int_events_aux_issues') }}