Test async background task

app.py

@@ -6,8 +6,9 @@ from huggingface_hub import HfApi, hf_hub_download, Repository
 from huggingface_hub.repocard import metadata_load
 import pandas as pd
 from matchmaking import *
-from background_task import init_matchmaking
+from background_task import init_matchmaking, run_background_loop
 from apscheduler.schedulers.background import BackgroundScheduler
+import asyncio
 
 
 DATASET_REPO_URL = "https://huggingface.co/datasets/CarlCochet/BotFightData"
@@ -22,9 +23,13 @@ matchmaking = Matchmaking()
 api = HfApi()
 
 
-scheduler = BackgroundScheduler()
-scheduler.add_job(func=init_matchmaking, trigger="interval", seconds=15000)
-scheduler.start()
+# scheduler = BackgroundScheduler()
+# scheduler.add_job(func=init_matchmaking, trigger="interval", seconds=15000)
+# scheduler.start()
+
+loop = asyncio.get_event_loop()
+loop.create_task(run_background_loop())
+loop.run_forever()
 
 
 def get_elo_data() -> pd.DataFrame:

background_task.py

@@ -1,5 +1,8 @@
 import os
+import time
 import random
+import asyncio
+import subprocess
 import pandas as pd
 from datetime import datetime
 from huggingface_hub import HfApi, Repository
@@ -64,12 +67,39 @@ class Matchmaking:
         while len(self.queue) > 1:
             model1 = self.queue.pop(0)
             model2 = self.queue.pop(self.find_n_closest_indexes(model1, 10))
-            result = match(model1, model2)
-            self.compute_elo(model1, model2, result)
+            match(model1, model2)
+        self.load_results()
+
+    def load_results(self):
+        """ Load the match history from the hub. """
+        repo.git_pull()
+        results = pd.read_csv(
+            "https://huggingface.co/datasets/huggingface-projects/temp-match-results/raw/main/results.csv"
+        )
+        # while len(results) < len(self.matches["model1"]):
+        #     time.sleep(60)
+        #     results = pd.read_csv(
+        #         "https://huggingface.co/datasets/huggingface-projects/temp-match-results/raw/main/results.csv"
+        #     )
+
+        for i, row in results.iterrows():
+            model1 = row["model1"].split("/")
+            model2 = row["model2"].split("/")
+            model1 = self.find_model(model1[0], model1[1])
+            model2 = self.find_model(model2[0], model2[1])
+            result = row["result"]
+            self.compute_elo(row["model1"], row["model2"], row["result"])
             self.matches["model1"].append(model1.name)
             self.matches["model2"].append(model2.name)
             self.matches["result"].append(result)
-            self.matches["datetime"].append(datetime.now().strftime("%Y-%m-%d %H:%M:%S.%f"))
+            self.matches["timestamp"].append(row["timestamp"])
+
+    def find_model(self, author, name):
+        """ Find a model in the models list. """
+        for model in self.models:
+            if model.author == author and model.name == name:
+                return model
+        return None
 
     def compute_elo(self, model1, model2, result):
         """ Compute the new elo for each model based on a match result. """
@@ -118,34 +148,27 @@ class Matchmaking:
         repo.push_to_hub(commit_message="Update ELO")
 
 
-def match(model1, model2) -> float:
+def match(model1, model2):
     """
-    !!! Current code is placeholder !!!
-    TODO: Launch a Unity process with the 2 models and get the result of the match
-
     :param model1: First Model object
     :param model2: Second Model object
     :return: match result (0: model1 lost, 0.5: draw, 1: model1 won)
     """
-    result = random.randint(0, 2) / 2
-
+    model1_id = model1.author + "/" + model1.name
+    model2_id = model2.author + "/" + model2.name
+    subprocess.run(["UnityEnvironment.exe", "-model1", model1_id, "-model2", model2_id])
     model1.games_played += 1
     model2.games_played += 1
-    return result
 
 
 def get_models_list() -> list:
     """
-    !!! Current code is placeholder !!!
-    TODO: Create a list of Model objects from the models found on the hub
-
     :return: list of Model objects
     """
     models = []
     models_names = []
     data = pd.read_csv(os.path.join(DATASET_REPO_URL, "resolve", "main", ELO_FILENAME))
-    # models_on_hub = api.list_models(filter=["reinforcement-learning", env, "stable-baselines3"])
-    models_on_hub = []
+    models_on_hub = api.list_models(filter=["reinforcement-learning", "ml-agents", "ML-Agents-SoccerTwos"])
     for i, row in data.iterrows():
         models.append(Model(row["author"], row["model"], row["elo"], row["games_played"]))
         models_names.append(row["model"])
@@ -163,6 +186,12 @@ def init_matchmaking():
     print("Matchmaking done ---", datetime.now().strftime("%Y-%m-%d %H:%M:%S.%f"))
 
 
+async def run_background_loop():
+    while True:
+        print("It's running!", datetime.now().strftime("%Y-%m-%d %H:%M:%S.%f"))
+        await asyncio.sleep(60)
+
+
 if __name__ == "__main__":
     print("It's running!")
     api = HfApi()