Fixing use of timestamp keyword in snowflake query (#1014)

* Changing the use of the timestamp keyword in snowflake query

* Updating unit tests

* Consistency funsies

data-workflows/activity/github_activity_model.py

@@ -9,7 +9,7 @@
 from pynamodb.attributes import UnicodeAttribute, NumberAttribute
 
 from utils.utils import get_current_timestamp, date_to_utc_timestamp_in_millis, datetime_to_utc_timestamp_in_millis
-from plugin.helpers import _get_cache, _get_repo_to_plugin_dict
+from plugin.helpers import _get_repo_to_plugin_dict
 
 
 LOGGER = logging.getLogger()
@@ -27,11 +27,11 @@ def __new__(cls, timestamp_formatter, type_identifier_formatter, query_projectio
         return github_activity_type
 
     LATEST = (datetime_to_utc_timestamp_in_millis, 'LATEST:{0}',
-              'repo AS name, to_timestamp(max(commit_author_date)) as latest_commit', 'name')
+              'repo AS name, TO_TIMESTAMP(MAX(commit_author_date)) AS latest_commit', 'name')
     MONTH = (date_to_utc_timestamp_in_millis, 'MONTH:{1:%Y%m}:{0}',
-             'repo AS name, date_trunc("month", to_date(commit_author_date)) as month, count(*) as commit_count',
+             'repo AS name, DATE_TRUNC("month", TO_DATE(commit_author_date)) AS month, COUNT(*) AS commit_count',
              'name, month')
-    TOTAL = (lambda timestamp: None, 'TOTAL:{0}', 'repo AS name, count(*) as commit_count', 'name')
+    TOTAL = (lambda timestamp: None, 'TOTAL:{0}', 'repo AS name, COUNT(*) AS commit_count', 'name')
 
     def format_to_timestamp(self, timestamp: datetime) -> Union[int, None]:
         return self.timestamp_formatter(timestamp)
@@ -43,7 +43,7 @@ def _create_subquery(self, plugins_by_earliest_ts: dict[str, datetime]) -> str:
         if self is GitHubActivityType.MONTH:
             return " OR ".join(
                 [
-                    f"repo = '{name}' AND to_timestamp(commit_author_date) >= "
+                    f"repo = '{name}' AND TO_TIMESTAMP(commit_author_date) >= "
                     f"{TIMESTAMP_FORMAT.format(ts.replace(day=1))}"
                     for name, ts in plugins_by_earliest_ts.items()
                 ]

data-workflows/activity/snowflake_adapter.py

@@ -38,16 +38,16 @@ def get_plugins_install_count_since_timestamp(plugins_by_earliest_ts: dict[str,
     query = f"""
             SELECT 
                 LOWER(file_project) AS name, 
-                {install_activity_type.get_query_timestamp_projection()} AS timestamp, 
+                {install_activity_type.get_query_timestamp_projection()} AS ts, 
                 COUNT(*) AS count
             FROM
                 imaging.pypi.labeled_downloads
             WHERE 
                 download_type = 'pip'
                 AND project_type = 'plugin'
                 AND ({_generate_subquery_by_type(plugins_by_earliest_ts, install_activity_type)})
-            GROUP BY name, timestamp
-            ORDER BY name, timestamp
+            GROUP BY name, ts
+            ORDER BY name, ts
             """
     LOGGER.info(f'Fetching data for granularity={install_activity_type.name}')
     return _mapped_query_results(query, 'PYPI', {}, _cursor_to_plugin_activity_mapper)

data-workflows/activity/tests/test_snowflake_adapter.py

@@ -42,16 +42,16 @@ def get_plugins_install_count_since_timestamp_query(projection, subquery):
     return f"""
             SELECT 
                 LOWER(file_project) AS name, 
-                {projection} AS timestamp, 
+                {projection} AS ts, 
                 COUNT(*) AS count
             FROM
                 imaging.pypi.labeled_downloads
             WHERE 
                 download_type = 'pip'
                 AND project_type = 'plugin'
                 AND ({subquery})
-            GROUP BY name, timestamp
-            ORDER BY name, timestamp
+            GROUP BY name, ts
+            ORDER BY name, ts
             """