Fix lifecycle contributor state accounting and add extra QA for sqlmesh metrics pipeline (#5177)

* feat: add first time contributor metrics and corresponding test cases

* feat: update contributor_activity_classification with test included for monthly artifact

* feat: add test for first time contributor monthly metrics

* refactor(metrics): rename new_contributors -> first_time_contributors and update related logic/tests

- Rename metric identifiers in SQL: lifecycle.sql now uses first_time_* (aliases, churn calc, and change_in fields)
- Update tests to reference first_time_contributors_* and change_in_first_time_contributors_*
- Remove obsolete bot_activity periodic test fixtures (monthly/quarterly/biannually)
- metrics_factories: remove early skip for new_contributors so first-time metrics are audited
- factory: adjust ignored incremental rule to first_time_contributors and tidy MetricsCycle error message formatting

* feat: add models and tests for last contributions to artifacts, collections, and projects

* feat: update contributor activity classification logic and add last contribution test data

---------

Co-authored-by: ken <[email protected]>
Co-authored-by: Carl Cervone <[email protected]>

Author: kenyiu

warehouse/metrics_tools/factory/factory.py

@@ -314,7 +314,7 @@ def queue_query(name: str):
                     parent_query = queries[table_name]["rendered_query"]
                     raise MetricsCycle(
                         textwrap.dedent(
-                            f"""Cycle from {name} to {table_name}: 
+                            f"""Cycle from {name} to {table_name}:
                          ---
                         {name}:
                         {rendered_query.sql(dialect="duckdb", pretty=True)}
@@ -744,7 +744,7 @@ def generate_time_aggregation_model_for_rendered_query(
             or "funding" in query_config["table_name"]
             or "releases" in query_config["table_name"]
             or "worldchain_users_aggregation" in query_config["table_name"]
-            or "new_contributors" in query_config["table_name"]
+            or "first_time_contributors" in query_config["table_name"]
         ):
             ignored_rules.append("incrementalmustdefinenogapsaudit")
 

warehouse/oso_sqlmesh/models/intermediate/events/github/first-last/int_last_contribution_to_artifact.sql

@@ -0,0 +1,27 @@
+/*
+TODO: This is a hack for now to fix performance issues with the contributor
+classifications. We should use some kind of factory for this in the future to
+get all dimensions
+*/
+MODEL (
+  name oso.int_last_contribution_to_artifact,
+  kind FULL,
+  partitioned_by (YEAR("time"), "event_source"),
+  grain (time, event_source, from_artifact_id, to_artifact_id),
+  audits (
+    has_at_least_n_rows(threshold := 0)
+  )
+);
+
+SELECT
+  MAX(time) AS time,
+  event_source,
+  from_artifact_id,
+  to_artifact_id
+FROM oso.int_first_of_event_from_artifact__github
+WHERE
+  event_type IN ('COMMIT_CODE', 'ISSUE_OPENED', 'PULL_REQUEST_OPENED', 'PULL_REQUEST_MERGED')
+GROUP BY
+  event_source,
+  from_artifact_id,
+  to_artifact_id

warehouse/oso_sqlmesh/models/intermediate/events/github/first-last/int_last_contribution_to_collection.sql

@@ -0,0 +1,30 @@
+/*
+TODO: This is a hack for now to fix performance issues with the contributor
+classifications. We should use some kind of factory for this in the future to
+get all dimensions
+*/
+MODEL (
+  name oso.int_last_contribution_to_collection,
+  kind FULL,
+  partitioned_by (YEAR("time"), "event_source"),
+  grain (time, event_source, from_artifact_id, to_collection_id),
+  tags (
+    'entity_category=collection'
+  ),
+  audits (
+    has_at_least_n_rows(threshold := 0)
+  )
+);
+
+SELECT
+  MAX(time) AS time,
+  last_contribution_to_project.event_source,
+  last_contribution_to_project.from_artifact_id,
+  projects_by_collection_v1.collection_id AS to_collection_id
+FROM oso.int_last_contribution_to_project AS last_contribution_to_project
+INNER JOIN oso.projects_by_collection_v1
+  ON last_contribution_to_project.to_project_id = projects_by_collection_v1.project_id
+GROUP BY
+  event_source,
+  from_artifact_id,
+  to_collection_id

warehouse/oso_sqlmesh/models/intermediate/events/github/first-last/int_last_contribution_to_project.sql

@@ -0,0 +1,30 @@
+/*
+TODO: This is a hack for now to fix performance issues with the contributor
+classifications. We should use some kind of factory for this in the future to
+get all dimensions
+*/
+MODEL (
+  name oso.int_last_contribution_to_project,
+  kind FULL,
+  partitioned_by (YEAR("time"), "event_source"),
+  grain (time, event_source, from_artifact_id, to_project_id),
+  tags (
+    'entity_category=project'
+  ),
+  audits (
+    has_at_least_n_rows(threshold := 0)
+  )
+);
+
+SELECT
+  MAX(last_contribution_to_artifact.time) AS time,
+  last_contribution_to_artifact.event_source,
+  last_contribution_to_artifact.from_artifact_id,
+  artifacts_by_project_v1.project_id AS to_project_id
+FROM oso.int_last_contribution_to_artifact AS last_contribution_to_artifact
+INNER JOIN oso.artifacts_by_project_v1
+  ON last_contribution_to_artifact.to_artifact_id = artifacts_by_project_v1.artifact_id
+GROUP BY
+  event_source,
+  from_artifact_id,
+  to_project_id

warehouse/oso_sqlmesh/models/metrics_factories.py

@@ -53,9 +53,6 @@ def no_gaps_audit_factory(config: MetricQueryConfig) -> MetricMetadataModifiers:
     if "worldchain_users_aggregation" in config["table_name"]:
         return {}
 
-    if "new_contributors" in config["table_name"]:
-        return {}
-
     if "data_category=blockchain" in config["additional_tags"]:
         options["ignore_before"] = constants.superchain_audit_start
         options["ignore_after"] = constants.superchain_audit_end
@@ -256,8 +253,8 @@ def add_project_and_collection_entity_category_tags(
             ),
             additional_tags=["data_category=code"],
         ),
-        "new_contributors": MetricQueryDef(
-            ref="code/new_contributors.sql",
+        "first_time_contributor": MetricQueryDef(
+            ref="code/first_time_contributor.sql",
             time_aggregations=[
                 "daily",
                 "weekly",
@@ -266,11 +263,10 @@ def add_project_and_collection_entity_category_tags(
                 # "biannually",
                 "yearly",
             ],
-            entity_types=["artifact", "project", "collection"],
             over_all_time=True,
             metadata=MetricMetadata(
-                display_name="New Contributors",
-                description="Metrics related to new GitHub contributors identified by author_association = 'FIRST_TIME_CONTRIBUTOR'",
+                display_name="First Time Contributors",
+                description="Metrics related to first time GitHub contributors",
             ),
             additional_tags=["data_category=code"],
         ),