refactor: remove functions responsible for handling team data

Author: hadiqur

src/main.py

@@ -59,64 +59,6 @@
 # }
 
 
-def get_copilot_team_date(gh: github_api_toolkit.github_interface, page: int) -> list:
-    """Gets a list of GitHub Teams with Copilot Data for a given API page.
-
-    Args:
-        gh (github_api_toolkit.github_interface): An instance of the github_interface class.
-        page (int): The page number of the API request.
-
-    Returns:
-        list: A list of GitHub Teams with Copilot Data.
-    """
-    copilot_teams = []
-
-    response = gh.get(f"/orgs/{org}/teams", params={"per_page": 100, "page": page})
-    teams = response.json()
-    for team in teams:
-        usage_data = gh.get(f"/orgs/{org}/team/{team['name']}/copilot/metrics")
-
-        if not isinstance(usage_data, Response):
-
-            # If the response is not a Response object, no copilot data is available for this team
-            # We can then skip this team
-
-            # We don't log this as an error, as it is expected and it'd be too noisy within logs
-
-            continue
-
-        # If the response has data, append the team to the list
-        # If there is no data, .json() will return an empty list
-        if usage_data.json():
-
-            team_name = team.get("name", "")
-            team_slug = team.get("slug", "")
-            team_description = team.get("description", "")
-            team_html_url = team.get("html_url", "")
-
-            logger.info(
-                "Team %s has Copilot data",
-                team_name,
-                extra={
-                    "team_name": team_name,
-                    "team_slug": team_slug,
-                    "team_description": team_description,
-                    "team_html_url": team_html_url,
-                },
-            )
-
-            copilot_teams.append(
-                {
-                    "name": team_name,
-                    "slug": team_slug,
-                    "description": team_description,
-                    "url": team_html_url,
-                }
-            )
-
-    return copilot_teams
-
-
 def get_and_update_historic_usage(
     s3: boto3.client, gh: github_api_toolkit.github_interface, write_data_locally: bool
 ) -> tuple:
@@ -176,104 +118,6 @@ def get_and_update_historic_usage(
     return sorted_historic_usage, dates_added
 
 
-def get_and_update_copilot_teams(
-    s3: boto3.client, gh: github_api_toolkit.github_interface, write_data_locally: bool
-) -> list:
-    """Get and update GitHub Teams with Copilot Data.
-
-    Args:
-        s3 (boto3.client): An S3 client.
-        gh (github_api_toolkit.github_interface): An instance of the github_interface class.
-        write_data_locally (bool): Whether to write data locally instead of to an S3 bucket.
-
-    Returns:
-        list: A list of GitHub Teams with Copilot Data.
-    """
-    logger.info("Getting GitHub Teams with Copilot Data")
-
-    copilot_teams = []
-
-    response = gh.get(f"/orgs/{org}/teams", params={"per_page": 100})
-
-    # Get the last page of teams
-    try:
-        last_page = int(response.links["last"]["url"].split("=")[-1])
-    except KeyError:
-        last_page = 1
-
-    for page in range(1, last_page + 1):
-        page_teams = get_copilot_team_date(gh, page)
-
-        copilot_teams = copilot_teams + page_teams
-
-    logger.info(
-        "Fetched GitHub Teams with Copilot Data",
-        extra={"no_teams": len(copilot_teams)},
-    )
-
-    if not write_data_locally:
-        update_s3_object(s3, BUCKET_NAME, "copilot_teams.json", copilot_teams)
-    else:
-        local_path = "output/copilot_teams.json"
-        os.makedirs("output", exist_ok=True)
-        with open(local_path, "w", encoding="utf-8") as f:
-            json.dump(copilot_teams, f, indent=4)
-        logger.info("Copilot teams data written locally to %s (S3 skipped)", local_path)
-
-    return copilot_teams
-
-
-def create_dictionary(
-    gh: github_api_toolkit.github_interface, copilot_teams: list, existing_team_history: list
-) -> list:
-    """Create a dictionary for quick lookup of existing team data using the `name` field.
-
-    Args:
-        gh (github_api_toolkit.github_interface): An instance of the github_interface class.
-        copilot_teams (list): List of teams with Copilot data.
-        existing_team_history (list): List of existing team history data.
-
-    Returns:
-        list: A list of dictionaries containing team data and their history.
-    """
-    existing_team_data_map = {
-        single_team["team"]["name"]: single_team for single_team in existing_team_history
-    }
-
-    # Iterate through identified teams
-    for team in copilot_teams:
-        team_name = team.get("name", "")
-        if not team_name:
-            logger.warning("Skipping team with no name")
-            continue
-
-        # Determine the last known date for the team
-        last_known_date = None
-        if team_name in existing_team_data_map:
-            existing_dates = [entry["date"] for entry in existing_team_data_map[team_name]["data"]]
-            if existing_dates:
-                last_known_date = max(existing_dates)  # Get the most recent date
-
-        # Assign the last known date to the `since` query parameter
-        query_params = {}
-        if last_known_date:
-            query_params["since"] = last_known_date
-
-        single_team_history = get_team_history(gh, team_name, query_params)
-        if not single_team_history:
-            logger.info("No new history found for team %s", team_name)
-            continue
-
-        # Append new data to the existing team history
-        new_team_data = single_team_history
-        if team_name in existing_team_data_map:
-            existing_team_data_map[team_name]["data"].extend(new_team_data)
-        else:
-            existing_team_data_map[team_name] = {"team": team, "data": new_team_data}
-
-    return list(existing_team_data_map.values())
-
-
 def update_s3_object(
     s3_client: boto3.client,
     bucket_name: str,
@@ -304,31 +148,6 @@ def update_s3_object(
         return False
 
 
-def get_team_history(
-    gh: github_api_toolkit.github_interface, team: str, query_params: Optional[dict] = None
-) -> list[dict]:
-    """Gets the team metrics Copilot data through the API.
-    Note - This endpoint will only return results for a given day if the team had
-    five or more members with active Copilot licenses on that day,
-    as evaluated at the end of that day.
-
-    Args:
-        gh (github_api_toolkit.github_interface): An instance of the github_interface class.
-        team (str): Team name.
-        query_params (dict): Additional query parameters for the API request.
-
-    Returns:
-        list[dict]: A team's GitHub Copilot metrics or None if an error occurs.
-    """
-    response = gh.get(f"/orgs/{org}/team/{team}/copilot/metrics", params=query_params)
-
-    if not isinstance(response, Response):
-        # If the response is not a Response object, no copilot data is available for this team
-        # We can return None which is then handled by the calling function
-        return None
-    return response.json()
-
-
 def get_dict_value(dictionary: dict, key: str) -> Any:
     """Gets a value from a dictionary and raises an exception if it is not found.
 
@@ -451,37 +270,6 @@ def handler(event: dict, context) -> str:  # pylint: disable=unused-argument, to
     # Copilot Usage Data (Historic)
     historic_usage, dates_added = get_and_update_historic_usage(s3, gh, write_data_locally)
 
-    # GitHub Teams with Copilot Data
-    copilot_teams = get_and_update_copilot_teams(s3, gh, write_data_locally)
-
-    logger.info("Getting history of each team identified previously")
-
-    # Retrieve existing team history from S3
-    try:
-        response = s3.get_object(Bucket=BUCKET_NAME, Key="teams_history.json")
-        existing_team_history = json.loads(response["Body"].read().decode("utf-8"))
-    except ClientError as e:
-        logger.warning("Error retrieving existing team history: %s", e)
-        existing_team_history = []
-
-    logger.info("Existing team history has %d entries", len(existing_team_history))
-
-    if not write_data_locally:
-        # Convert to dictionary for quick lookup
-        updated_team_history = create_dictionary(gh, copilot_teams, existing_team_history)
-
-        # Write updated team history to S3
-        # This line isn't covered by tests as it's painful to get to.
-        # The function itself is tested though.
-        update_s3_object(s3, BUCKET_NAME, "teams_history.json", updated_team_history)
-    else:
-        local_path = "output/teams_history.json"
-        os.makedirs("output", exist_ok=True)
-        updated_team_history = create_dictionary(gh, copilot_teams, existing_team_history)
-        with open(local_path, "w", encoding="utf-8") as f:
-            json.dump(updated_team_history, f, indent=4)
-        logger.info("Team history written locally to %s (S3 skipped)", local_path)
-
     logger.info(
         "Process complete",
         extra={
@@ -490,14 +278,13 @@ def handler(event: dict, context) -> str:  # pylint: disable=unused-argument, to
             "dates_added": dates_added,
             "no_dates_before": len(historic_usage) - len(dates_added),
             "no_dates_after": len(historic_usage),
-            "no_copilot_teams": len(copilot_teams),
         },
     )
 
     return "Github Data logging is now complete."
 
 
-# # Dev Only
-# # Uncomment the following line to run the script locally
-if __name__ == "__main__":
-    handler(None, None)
+# Dev Only
+# Uncomment the following line to run the script locally
+# if __name__ == "__main__":
+#     handler(None, None)