Merge pull request #74 from srvanrell/stop-github-workflow-no-updates

Stop GitHub workflow no updates

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ranking_table_tennis"
-version = "2024.6.17"
+version = "2024.6.18"
 description = "A ranking table tennis system"
 readme = "README.md"
 authors = [ "Sebastian Vanrell srvanrell_gmail_com" ]

ranking_table_tennis/helpers/__init__.py

@@ -8,6 +8,7 @@
     save_ranking_sheet,
     save_raw_ranking,
 )
+from ranking_table_tennis.helpers.github import no_updates_stop_workflow
 from ranking_table_tennis.helpers.gspread import publish_to_web
 from ranking_table_tennis.helpers.initial_rating import print_rating_context
 from ranking_table_tennis.helpers.pickle import load_from_pickle, save_to_pickle

ranking_table_tennis/helpers/github.py

@@ -0,0 +1,25 @@
+import logging
+import os
+
+from ranking_table_tennis.helpers.gspread import days_since_last_update
+
+logger = logging.getLogger(__name__)
+
+
+def no_updates_stop_workflow(spreadsheet_id, max_days_since_last_update=3):
+    """If spreasheet_id has not been modified recently GITHUB_OUTPUT env var will stop workflow"""
+
+    days_since_update = days_since_last_update(spreadsheet_id)
+    stop_workflow = days_since_update > max_days_since_last_update
+    logger.info(
+        "%s days since last update. Should stop workflow? %s", days_since_update, stop_workflow
+    )
+
+    GITHUB_OUTPUT_VAR_NAME = "GITHUB_OUTPUT"
+    var_value = f"stop_workflow={str(stop_workflow).lower()}"  # true or false
+
+    logger.debug("Set env var '%s': '%s'", GITHUB_OUTPUT_VAR_NAME, var_value)
+    os.environ[GITHUB_OUTPUT_VAR_NAME] = var_value
+
+    if stop_workflow:
+        exit(0)

ranking_table_tennis/helpers/gspread.py

@@ -1,6 +1,7 @@
 import json
 import logging
 import os
+from datetime import datetime, timezone
 from typing import List
 
 import gspread
@@ -104,6 +105,21 @@ def publish_to_web(tid: str, show_on_web=False) -> None:
             create_n_tour_sheet(spreadsheet_id, tid)
 
 
+def days_since_last_update(spreadsheet_id) -> str:
+    try:
+        gc = _get_gc()
+        spreadsheet = gc.open_by_key(spreadsheet_id)
+        last_update_time = spreadsheet.get_lastUpdateTime()
+    except PermissionError:
+        logger.warn("!! Permission Error. FAIL TO GET last updated time @ '%s'", spreadsheet_id)
+
+    # Parse the ISO 8601 datetime string into a datetime at UTC
+    last_update_utc = datetime.fromisoformat(last_update_time[:-1]).replace(tzinfo=timezone.utc)
+    difference = datetime.now(timezone.utc) - last_update_utc
+
+    return difference.days
+
+
 def _in_colab() -> bool:
     # Verify if it is running on colab
     try:
@@ -123,11 +139,11 @@ def _get_gc() -> gspread.Client:
         github_secret_name = "GCP_SA_KEY"
         credentials_str = os.getenv(github_secret_name)
         if credentials_str:
-            logger.debug("Loading GCP credentials from %s", github_secret_name)
+            logger.debug("Loading GCP SA credentials from %s", github_secret_name)
             credentials_dict = json.loads(credentials_str)
             gc = gspread.service_account_from_dict(credentials_dict)
         else:
-            logger.debug("Loading GCP credentials from standard path")
+            logger.debug("Loading GCP SA credentials from standard path")
             gc = gspread.service_account()
     except FileNotFoundError:
         logger.warn(
@@ -136,6 +152,7 @@ def _get_gc() -> gspread.Client:
 
     try:
         if gc is None:
+            logger.debug("Loading GCP End User credentials from standard path")
             gc = gspread.oauth()
     except FileNotFoundError:
         logger.warn("!! The end user .json key file has not been configured. Upload might fail.")

ranking_table_tennis/preprocess.py

@@ -28,22 +28,18 @@ def main(offline=True, assume_yes=False, config_initial_date="220101"):
     xlsx_file = cfg.io.data_folder + cfg.io.xlsx.tournaments_filename
 
     if not offline:
+        # Stop preprocess if there were no recent updates on the spreadsheet
+        helpers.no_updates_stop_workflow(cfg.io.tournaments_spreadsheet_id)
+
         if assume_yes:
             retrieve = "yes"
         else:
             retrieve = input("\nDo you want to retrieve online sheet [Y/n]? ")
+
         if retrieve.lower() != "n":
             logger.info("Downloading and saving '%s'" % xlsx_file)
             request.urlretrieve(cfg.io.tournaments_gdrive, xlsx_file)
 
-    # Help to stop the workflow if there are no updates on the file
-    # TODO exit if there are no updates.
-    new_data_available = True
-    if new_data_available:
-        print("::set-output name=stop_workflow::false")
-    else:
-        print("::set-output name=stop_workflow::true")
-
     # Loading all tournament data
     tournaments = helpers.load_tournaments_sheets()
 

ranking_table_tennis/preprocess_unattended.py

@@ -25,17 +25,12 @@ def main(config_initial_date="220101"):
     ConfigManager().set_current_config(date=config_initial_date)
     cfg = ConfigManager().current_config
 
-    xlsx_file = cfg.io.data_folder + cfg.io.xlsx.tournaments_filename
+    # Stop preprocess if there were no recent updates on the spreadsheet
+    helpers.no_updates_stop_workflow(cfg.io.tournaments_spreadsheet_id)
 
+    xlsx_file = cfg.io.data_folder + cfg.io.xlsx.tournaments_filename
     logger.info("Downloading and saving '%s'" % xlsx_file)
     request.urlretrieve(cfg.io.tournaments_gdrive, xlsx_file)
-    # Help to stop the workflow if there are no updates on the file
-    # TODO exit if there are no updates.
-    new_data_available = True
-    if new_data_available:
-        print("::set-output name=stop_workflow::false")
-    else:
-        print("::set-output name=stop_workflow::true")
 
     # Loading all tournament data
     tournaments = helpers.load_tournaments_sheets()