fix: analytics package issues (#4352)

* fix: renamed rowsCount to rowCount (#4351)

* fix: switched ga result filtering from results to dimension filters (#4351)

* chore: bumped analytics package version (#4351)

* chore: updated analytics package docstrings (#4351)

* chore: remove test csv outputs (#4351)

---------

Co-authored-by: hunterckx <118154470+hunterckx@users.noreply.github.com>

Author: Jonah Paten

analytics/analytics_package/analytics/api.py

@@ -160,7 +160,7 @@ def get_metrics_by_dimensions_v4_style(service, metrics, dimensions, property, s
 	while has_more:
 		result = service.properties().runReport(property=property, body=params).execute()
 		if rows_left is None:
-			rows_left = result.get("rowsCount", 0)
+			rows_left = result.get("rowCount", 0)
 		page_row_count = len(result["rows"]) if "rows" in result else 0
 		has_more = page_row_count > 0
 		if has_more:
@@ -172,7 +172,7 @@ def get_metrics_by_dimensions_v4_style(service, metrics, dimensions, property, s
 				offset += max_results
 				params["offset"] = offset
 
-	df =  v4_results_to_df(results, dimensions, metrics)
+	df = v4_results_to_df(results, dimensions, metrics)
 
 	return df
 

analytics/analytics_package/analytics/fields.py

@@ -3,24 +3,25 @@
 METRIC_TOTAL_USERS = 'totalUsers'
 METRIC_PAGE_VIEW = 'screenPageViews'
 
+# Event Names
+EVENT_BUILTIN_CLICK = "click"
+EVENT_CUSTOM_CLICK = "outbound_link_clicked"
+EVENT_PAGE_VIEW = "page_view"
+
 # DIMENSIONS
 DIMENSION_PAGE_PATH = {
     'id': 'pagePath',
     'alias': 'page_path',
-    'remove_matches': None,
 }
 DIMENSION_BUILTIN_URL = {
     'id': 'linkUrl',
     'alias': 'builtin_url',
-    'remove_matches': r"\s*",
 }
 DIMENSION_EVENT_NAME = {
     'id': 'eventName',
     'alias': 'event_name',
-    'remove_matches': None,
 }
 DIMENSION_CUSTOM_URL = {
     'id': 'customEvent:click_url',
     'alias': 'outbound_url',
-    'remove_matches': r"\(not set\)",
 }

analytics/analytics_package/analytics/sheets_elements.py

@@ -5,31 +5,20 @@
 from urllib.parse import urlparse
 import datetime as dt
 
-def get_flat_data_df(metrics, dimensions, remove_matches=None, **other_params):
+def get_flat_data_df(metrics, dimensions, **other_params):
     """
     Get a df from the Analytics API with a flat structure (no multiindex).
 
     :param analytics_params: the parameters for the Analytics API, including authentication and property ids
     :param metrics: the metrics to get
     :param dimensions: the dimensions to get
-    :param remove_matches: a list of regex patterns or None elements to remove from each dimension. 
-        Each regex or None element should correspond with an element of dimensions and remove_matches must be the same length as dimensions. 
-        If the value is None, no patterns are removed, defaults to None.
-
     :return: a DataFrame with the data from the Analytics API
     """
-    if remove_matches is not None:
-        assert len(remove_matches) == len(dimensions)
-
     df = get_data_df(
         metrics,
         [dimension["id"] for dimension in dimensions],
         **other_params,
     )
-    if remove_matches is not None:
-        for i, match in enumerate([dimension["remove_matches"] for dimension in dimensions]):
-            if match is not None:
-                df = df.loc[~df.index.get_level_values(i).str.fullmatch(match)]
     return df.reset_index().rename(columns=get_rename_dict(dimensions)).copy()
 
 def get_rename_dict(dimensions):
@@ -41,26 +30,27 @@ def get_rename_dict(dimensions):
 def get_outbound_links_df(analytics_params):
     """
     Get a DF with outbound links from the Analytics API. Merges the builtin and custom events for outbound links.
+    analytics_params cannot currently include a dimension_filter
 
     :param analytics_params: the parameters for the Analytics API, including authentication and property ids
     :return: a DataFrame with the outbound links from the Analytics API
     """
     pd.set_option('future.no_silent_downcasting', True)
+    assert "dimension_filter" not in analytics_params
     # Get the builtin "Click" event
     df_builtin_links = get_flat_data_df(
         [METRIC_EVENT_COUNT, METRIC_TOTAL_USERS],
         [DIMENSION_PAGE_PATH, DIMENSION_BUILTIN_URL, DIMENSION_EVENT_NAME],
-        remove_matches=[None, r"\s*", None],
+        dimension_filter=f"eventName=={EVENT_BUILTIN_CLICK}",
         **analytics_params,
     ).groupby(
         [DIMENSION_PAGE_PATH["alias"], DIMENSION_BUILTIN_URL["alias"]]
     ).sum().reset_index()
-
     # Get the custom "outbound_link_click" event
     df_custom_links = get_flat_data_df(
         [METRIC_EVENT_COUNT, METRIC_TOTAL_USERS],
         [DIMENSION_EVENT_NAME, DIMENSION_CUSTOM_URL, DIMENSION_PAGE_PATH], 
-        remove_matches=[DIMENSION_EVENT_NAME["remove_matches"], r"\(not set\)", None],
+        dimension_filter=f"eventName=={EVENT_CUSTOM_CLICK}",
         **analytics_params, 
     ).groupby(
         [DIMENSION_PAGE_PATH["alias"], DIMENSION_CUSTOM_URL["alias"]]
@@ -153,11 +143,12 @@ def get_page_views_df(analytics_params):
     :param analytics_params: the parameters for the Analytics API, including authentication and property ids
     :return: a DataFrame with the page views from the Analytics API
     """
+    assert "dimension_filter" not in analytics_params
     df_response = get_flat_data_df(
         [METRIC_EVENT_COUNT, METRIC_TOTAL_USERS, METRIC_PAGE_VIEW],
         [DIMENSION_PAGE_PATH, DIMENSION_EVENT_NAME],
-        dimension_filter="eventName==page_view",
         **analytics_params,
+        dimension_filter=f"eventName=={EVENT_PAGE_VIEW}",
     ).rename(
         columns={
             DIMENSION_PAGE_PATH["alias"]: "Page Path",

analytics/analytics_package/setup.py

@@ -2,7 +2,7 @@
 
 setup(
 	name="analytics",
-	version="3.3.0",
+	version="3.3.1",
 	packages=["analytics"],
 	install_requires=["matplotlib", "pandas", "numpy", "google-auth-oauthlib", "google-api-python-client", "gspread", "gspread-formatting"],
 )