added mode sql code for fetching github commits and updated the model

data-workflows/activity/github_activity_model.py

@@ -35,21 +35,21 @@ def format_to_timestamp(self, timestamp: datetime) -> Union[int, None]:
     def format_to_type_identifier(self, identifier: str) -> str:
         return self.type_identifier_formatter.format(identifier)
 
-    def get_query_timestamp_projection(self) -> str:
+    def get_query_projection(self) -> str:
         if self is GitHubActivityType.LATEST:
-            return '1'
-        elif self is GitHubActivityType.TOTAL:
-            return '2'
+            return '1, max(commit_author_date) as latest_commit'
+        elif self is GitHubActivityType.MONTH:
+            return 'date_trunc('"month"', to_date(commit_author_date)) as month, count(*) as num_commit'
         else:
-            return f"DATE_TRUNC('{self.name}', timestamp)"
+            return '1, count(*) as num_commits'
 
 
 class GitHubActivity(Model):
     class Meta:
+        host = os.getenv('LOCAL_DYNAMO_HOST')
         prefix = os.getenv('STACK_NAME')
         region = os.getenv('AWS_REGION')
         table_name = f'{prefix}-github-activity'
-        region = region
 
     plugin_name = UnicodeAttribute(hash_key=True)
     type_identifier = UnicodeAttribute(range_key=True)

data-workflows/activity/snowflake_adapter.py

@@ -54,6 +54,25 @@ def get_plugins_install_count_since_timestamp(plugins_by_earliest_ts: dict[str,
     return _mapped_query_results(query, 'PYPI', {}, _cursor_to_plugin_activity_mapper)
 
 
+def get_plugins_commit_count_since_timestamp(plugins_by_earliest_ts: dict[str, datetime],
+                                             github_activity_type: GitHubActivityType) -> dict[str, List]:
+    # TODO: fill in (...) for the logic to get the commit count since a specific starting point for each plugin
+    query = f"""
+            SELECT 
+                LOWER(file_project) AS plugin, 
+                {github_activity_type.get_query_projection()}
+            FROM
+                imaging.github.commits
+            WHERE 
+                repo_type = 'plugin'
+                AND (...)
+            GROUP BY 1, 2
+            ORDER BY 1, 2
+            """
+    LOGGER.info(f'Fetching data for granularity={github_activity_type.name}')
+    return _mapped_query_results(query, 'GITHUB', {}, _cursor_to_plugin_activity_mapper)
+
+
 def get_plugins_with_commits_in_window(start_millis: int, end_millis: int) -> dict[str, datetime]:
     query = f"""
             SELECT