This will be performed by ALL the games returning a JSON object for feedback instead of a strong
def run_single_game_with_feedback(cls, league, custom_rewards=None):
game = cls(league, verbose=True)
results = game.play_game(custom_rewards)
game_feedback = "\n".join(game.game_feedback)
player_feedback = "\n".join(game.player_feedback)
return {
"results": results,
"feedback": game_feedback, #CONVERT FROM MD FORMATTED STRONG TO JSON OBJ
"player_feedback": player_feedback #CONVERT FROM MD FORMATTED STRONG TO JSON OBJ
}
player_feedback and game_feedback are basically the same but the player feedback includes stuff that the player printed by using self.add_feedback.
Thse files will need to be updated:
backend/docker_simulation.py has a validation JSON schema that will also need to be changed
existing games: greedy_pig and prisoners_dilemma will need to be changed to return JSON
frontend/src/AgentGames/Utilities/MarkdownFeedback.jsx will need to be replaced with something that prints formatted json
Specific games could then have their own react component to display the results
Once this is implemented we will no longer need docker_results.json. The feedback can be saved to the DB using https://sqlite.org/jsonb.html
This will be performed by ALL the games returning a JSON object for feedback instead of a strong
player_feedback and game_feedback are basically the same but the player feedback includes stuff that the player printed by using self.add_feedback.
Thse files will need to be updated:
backend/docker_simulation.pyhas a validation JSON schema that will also need to be changed existing games:greedy_pigandprisoners_dilemmawill need to be changed to return JSONfrontend/src/AgentGames/Utilities/MarkdownFeedback.jsxwill need to be replaced with something that prints formatted jsonSpecific games could then have their own react component to display the results
Once this is implemented we will no longer need docker_results.json. The feedback can be saved to the DB using https://sqlite.org/jsonb.html