Django init

.gitignore

@@ -21,3 +21,6 @@
 npm-debug.log*
 yarn-debug.log*
 yarn-error.log*
+
+migrations/
+db.sqlite3

README.md

@@ -1,5 +1,40 @@
 **NOTE: This project is under final tesing. The one in the repo only supports the visualization of some sampled data. The full version will be available soon!**
 
+# Server Setup
+Install dependencies:
+```
+pip install -r requirements.txt
+```
+Migrate the databases:
+```
+cd server
+python manage.py makemigrations
+python manage.py migrate
+```
+Run server:
+```
+python manage.py runserver
+```
+The default URL is [http://127.0.0.1:8000/](http://127.0.0.1:8000/)
+
+# REST API
+The definitions of the fields are as follows:
+*   `eval_num`: Integer. The number of evaluation times.
+*   `name`: String. The name of the environment.
+*   `agent0`: String. Model name.
+*   `agent1`: String. Model name.
+*   `win`: Boolean. True if model in the first seat wins.
+*   `payoff`: Float. The payoff of the agent in the first seat.
+*   `index`: Integer. The index of the game of the same environent and same agent. It is in the range \[0, eval\_num-1\]
+
+| type | Resource                  |  Parameters                                          |  Description                                                                                                       |
+|------|---------------------------|------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------|
+| GET  | tournament/launch         | `eval\_num`, `name`                                  | Launch tournment on the game. Each pair of models will play `eval\_num` times. Results will be saved in database.  |
+| GET  | tournament/query\_game    | `name`, `index`, `agent0`, `agent1`, `win`, `payoff` | Query the games with the given parameters                                                                          |
+| GET  | tournament/query\_payoff  | `name`, `agent0`, `agent1`, `payoff`                 | Query the payoffs with the given parameters                                                                        |
+| GET  | tournament/replay         | `name`, `agent0`, `agent1`, `index`                  | Return the replay data (only support Leduc Holdem for now)                                                         |
+
+# AAA
 This project was bootstrapped with [Create React App](https://github.com/facebook/create-react-app).
 
 ## Available Scripts

requirements.txt

@@ -0,0 +1,4 @@
+rlcard
+Django
+tqdm
+tensorflow==1.14

server/manage.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+"""Django's command-line utility for administrative tasks."""
+import os
+import sys
+
+
+def main():
+    os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'server.settings')
+    try:
+        from django.core.management import execute_from_command_line
+    except ImportError as exc:
+        raise ImportError(
+            "Couldn't import Django. Are you sure it's installed and "
+            "available on your PYTHONPATH environment variable? Did you "
+            "forget to activate a virtual environment?"
+        ) from exc
+    execute_from_command_line(sys.argv)
+
+
+if __name__ == '__main__':
+    main()

server/server/settings.py

@@ -0,0 +1,121 @@
+"""
+Django settings for server project.
+
+Generated by 'django-admin startproject' using Django 2.2.12.
+
+For more information on this file, see
+https://docs.djangoproject.com/en/2.2/topics/settings/
+
+For the full list of settings and their values, see
+https://docs.djangoproject.com/en/2.2/ref/settings/
+"""
+
+import os
+
+# Build paths inside the project like this: os.path.join(BASE_DIR, ...)
+BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+
+
+# Quick-start development settings - unsuitable for production
+# See https://docs.djangoproject.com/en/2.2/howto/deployment/checklist/
+
+# SECURITY WARNING: keep the secret key used in production secret!
+SECRET_KEY = '-t@mf4fi)gfxzv5lm8qkg)*5u^brj--y*ul2&ryqdem(xin8(!'
+
+# SECURITY WARNING: don't run with debug turned on in production!
+DEBUG = True
+
+ALLOWED_HOSTS = []
+
+
+# Application definition
+
+INSTALLED_APPS = [
+    'django.contrib.admin',
+    'django.contrib.auth',
+    'django.contrib.contenttypes',
+    'django.contrib.sessions',
+    'django.contrib.messages',
+    'django.contrib.staticfiles',
+    'tournament',
+]
+
+MIDDLEWARE = [
+    'django.middleware.security.SecurityMiddleware',
+    'django.contrib.sessions.middleware.SessionMiddleware',
+    'django.middleware.common.CommonMiddleware',
+    'django.middleware.csrf.CsrfViewMiddleware',
+    'django.contrib.auth.middleware.AuthenticationMiddleware',
+    'django.contrib.messages.middleware.MessageMiddleware',
+    'django.middleware.clickjacking.XFrameOptionsMiddleware',
+]
+
+ROOT_URLCONF = 'server.urls'
+
+TEMPLATES = [
+    {
+        'BACKEND': 'django.template.backends.django.DjangoTemplates',
+        'DIRS': [],
+        'APP_DIRS': True,
+        'OPTIONS': {
+            'context_processors': [
+                'django.template.context_processors.debug',
+                'django.template.context_processors.request',
+                'django.contrib.auth.context_processors.auth',
+                'django.contrib.messages.context_processors.messages',
+            ],
+        },
+    },
+]
+
+WSGI_APPLICATION = 'server.wsgi.application'
+
+
+# Database
+# https://docs.djangoproject.com/en/2.2/ref/settings/#databases
+
+DATABASES = {
+    'default': {
+        'ENGINE': 'django.db.backends.sqlite3',
+        'NAME': os.path.join(BASE_DIR, 'db.sqlite3'),
+    }
+}
+
+
+# Password validation
+# https://docs.djangoproject.com/en/2.2/ref/settings/#auth-password-validators
+
+AUTH_PASSWORD_VALIDATORS = [
+    {
+        'NAME': 'django.contrib.auth.password_validation.UserAttributeSimilarityValidator',
+    },
+    {
+        'NAME': 'django.contrib.auth.password_validation.MinimumLengthValidator',
+    },
+    {
+        'NAME': 'django.contrib.auth.password_validation.CommonPasswordValidator',
+    },
+    {
+        'NAME': 'django.contrib.auth.password_validation.NumericPasswordValidator',
+    },
+]
+
+
+# Internationalization
+# https://docs.djangoproject.com/en/2.2/topics/i18n/
+
+LANGUAGE_CODE = 'en-us'
+
+TIME_ZONE = 'UTC'
+
+USE_I18N = True
+
+USE_L10N = True
+
+USE_TZ = True
+
+
+# Static files (CSS, JavaScript, Images)
+# https://docs.djangoproject.com/en/2.2/howto/static-files/
+
+STATIC_URL = '/static/'

server/server/urls.py

@@ -0,0 +1,22 @@
+"""server URL Configuration
+
+The `urlpatterns` list routes URLs to views. For more information please see:
+    https://docs.djangoproject.com/en/2.2/topics/http/urls/
+Examples:
+Function views
+    1. Add an import:  from my_app import views
+    2. Add a URL to urlpatterns:  path('', views.home, name='home')
+Class-based views
+    1. Add an import:  from other_app.views import Home
+    2. Add a URL to urlpatterns:  path('', Home.as_view(), name='home')
+Including another URLconf
+    1. Import the include() function: from django.urls import include, path
+    2. Add a URL to urlpatterns:  path('blog/', include('blog.urls'))
+"""
+from django.contrib import admin
+from django.urls import include, path
+
+urlpatterns = [
+    path('tournament/', include('tournament.urls')),
+    path('admin/', admin.site.urls),
+]

server/server/wsgi.py

@@ -0,0 +1,16 @@
+"""
+WSGI config for server project.
+
+It exposes the WSGI callable as a module-level variable named ``application``.
+
+For more information on this file, see
+https://docs.djangoproject.com/en/2.2/howto/deployment/wsgi/
+"""
+
+import os
+
+from django.core.wsgi import get_wsgi_application
+
+os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'server.settings')
+
+application = get_wsgi_application()

server/tournament/admin.py

@@ -0,0 +1,3 @@
+from django.contrib import admin
+
+# Register your models here.

server/tournament/apps.py

@@ -0,0 +1,5 @@
+from django.apps import AppConfig
+
+
+class TournamentConfig(AppConfig):
+    name = 'tournament'

server/tournament/models.py

@@ -0,0 +1,37 @@
+from django.db import models
+
+class Game(models.Model):
+    # The name of the game
+    name = models.CharField(max_length=100)
+
+    # The ID of repeated games
+    index = models.CharField(max_length=100)
+
+    # The first agent
+    agent0 = models.CharField(max_length=100)
+
+    # The second agent
+    agent1 = models.CharField(max_length=100)
+
+    # Whether the first agent wins
+    win = models.BooleanField()
+
+    # The payoff of the first agent
+    payoff = models.FloatField()
+
+    # The JSON file
+    replay = models.TextField(blank=True)
+
+class Payoff(models.Model):
+    # The name of the game
+    name = models.CharField(max_length=100)
+
+    # The first agent
+    agent0 = models.CharField(max_length=100)
+
+    # The second agent
+    agent1 = models.CharField(max_length=100)
+ 
+    # The average payoff of the first agent
+    payoff = models.FloatField()
+

server/tournament/rlcard_wrap/__init__.py

@@ -0,0 +1,15 @@
+import rlcard
+from .leduc_holdem_random_model import LeducHoldemRandomModelSpec
+
+
+# Register Leduc Holdem Random Model
+rlcard.models.registration.model_registry.model_specs['leduc-holdem-random'] = LeducHoldemRandomModelSpec()
+
+# The models we are concerned
+MODEL_IDS = {}
+MODEL_IDS['leduc-holdem'] = [
+        'leduc-holdem-random',
+        'leduc-holdem-cfr',
+        'leduc-holdem-rule-v1',
+        ]
+

server/tournament/rlcard_wrap/leduc_holdem_random_model.py

@@ -0,0 +1,46 @@
+# A wrap for rlcard
+# Here, we include a random model as the default baseline
+import rlcard
+from rlcard.agents import RandomAgent
+from rlcard.models.model import Model
+
+class LeducHoldemRandomModelSpec(object):
+    def __init__(self):
+        self.model_id = 'leduc-holdem-random'
+        self._entry_point = LeducHoldemRandomModel
+
+    def load(self):
+        model = self._entry_point()
+        return model
+
+class LeducHoldemRandomModel(Model):
+    ''' A random model
+    '''
+
+    def __init__(self):
+        ''' Load random model
+        '''
+        env = rlcard.make('leduc-holdem')
+        self.agent = RandomAgent(action_num=env.action_num)
+        self.player_num = env.player_num
+
+    @property
+    def agents(self):
+        ''' Get a list of agents for each position in a the game
+
+        Returns:
+            agents (list): A list of agents
+
+        Note: Each agent should be just like RL agent with step and eval_step
+              functioning well.
+        '''
+        return [self.agent for _ in range(self.player_num)]
+
+    @property
+    def use_raw(self):
+        ''' Indicate whether use raw state and action
+
+        Returns:
+            use_raw (boolean): True if using raw state and action
+        '''
+        return False

server/tournament/tests.py

@@ -0,0 +1,3 @@
+from django.test import TestCase
+
+# Create your tests here.

server/tournament/tournament.py

@@ -0,0 +1,117 @@
+import os
+import json
+from tqdm import tqdm
+import numpy as np
+
+from .rlcard_wrap import rlcard
+
+class Tournament(object):
+    
+    def __init__(self, game, model_ids, evaluate_num=100):
+        """ Defalt for two player games
+            For Dou Dizhu, the two peasants use the same model
+        """
+        self.game = game
+        self.model_ids = model_ids
+        self.evaluate_num = evaluate_num
+        # Load the models
+        self.models = [rlcard.models.load(model_id) for model_id in model_ids]
+
+    def launch(self):
+        """ Currently for two-player game only
+        """
+        model_num = len(self.model_ids)
+        games_data = []
+        payoffs_data = []
+        for i in range(model_num):
+            for j in range(model_num):
+                if j == i:
+                    continue
+                print(self.game, '-', self.model_ids[i], 'VS', self.model_ids[j])
+                data, payoffs, wins = tournament(self.game, [self.models[i].agents[0], self.models[j].agents[1]], self.evaluate_num)
+                mean_payoff = np.mean(payoffs)
+                print('Average payoff:', mean_payoff)
+                print()
+
+                for k in range(len(data)):
+                    game_data = {}
+                    game_data['name'] = self.game
+                    game_data['index'] = k
+                    game_data['agent0'] = self.model_ids[i]
+                    game_data['agent1'] = self.model_ids[j]
+                    game_data['win'] = wins[k]
+                    game_data['replay'] = data[k]
+                    game_data['payoff'] = payoffs[k]
+
+                    games_data.append(game_data)
+
+                payoff_data = {}
+                payoff_data['name'] = self.game
+                payoff_data['agent0'] = self.model_ids[i]
+                payoff_data['agent1'] = self.model_ids[j]
+                payoff_data['payoff'] = mean_payoff
+                payoffs_data.append(payoff_data)
+        return games_data, payoffs_data
+
+def tournament(game, agents, num):
+    env = rlcard.make(game, config={'allow_raw_data': True})
+    env.set_agents(agents)
+    payoffs = []
+    json_data = []
+    wins = []
+    for _ in tqdm(range(num)):
+        data = {}
+        data['playerInfo'] = [{'id': i, 'index': i} for i in range(env.player_num)]
+        state, player_id = env.reset()
+        perfect = env.get_perfect_information()
+        data['initHands'] = perfect['hand_cards']
+        data['moveHistory'] = [[]]
+        while not env.is_over():
+            action, probs = env.agents[player_id].eval_step(state)
+            history = {}
+            history['playerIdx'] = player_id
+            if env.agents[player_id].use_raw:
+                history['move'] = action
+            else:
+                history['move'] = env._decode_action(action)
+
+            probabilities = []
+            for i, a in enumerate(env.actions):
+                if len(probs) == 0:
+                    p = -2
+                elif a in state['raw_legal_actions']:
+                    p = probs[i]
+                else:
+                    p = -1
+                probabilities.append({'move':a, 'probability': p})
+            history['probabilities'] = probabilities
+            data['moveHistory'][0].append(history)
+            state, player_id = env.step(action, env.agents[player_id].use_raw)
+        perfect = env.get_perfect_information()
+        data['publicCard'] = perfect['public_card']
+        data = json.dumps(data)
+        #data = json.dumps(data, indent=2, sort_keys=True)
+        json_data.append(data)
+        if env.get_payoffs()[0] > 0:
+            wins.append(True)
+        else:
+            wins.append(False)
+        payoffs.append(env.get_payoffs()[0])
+    return json_data, payoffs, wins
+
+if __name__=='__main__':
+    game = 'leduc-holdem'
+    model_ids = ['leduc-holdem-random', 'leduc-holdem-rule-v1', 'leduc-holdem-cfr']
+    t = Tournament(game, model_ids)
+    games_data = t.launch()
+    print(len(games_data))
+    print(games_data[0])
+    #root_path = './models'
+    #agent1 = LeducHoldemDQNModel1(root_path)
+    #agent2 = LeducHoldemRandomModel(root_path)
+    #agent3 = LeducHoldemRuleModel()
+    #agent4 = LeducHoldemCFRModel(root_path)
+    #agent5 = LeducHoldemDQNModel2(root_path)
+    #t = Tournament(agent1, agent2, agent3, agent4, agent5, 'leduc-holdem')
+    ##t.competition()
+    #t.evaluate()

server/tournament/urls.py

@@ -0,0 +1,10 @@
+from django.urls import path
+
+from . import views
+
+urlpatterns = [
+        path('replay', views.replay, name='replay'),
+        path('launch', views.launch, name='launch'),
+        path('query_payoff', views.query_payoff, name='query_payoff'),
+        path('query_game', views.query_game, name='query_game'),
+        ]

server/tournament/views.py

@@ -0,0 +1,60 @@
+from django.shortcuts import render
+from django.http import HttpResponse
+from django.db import transaction
+from django.core import serializers
+
+from .models import Game, Payoff
+
+from .rlcard_wrap import rlcard, MODEL_IDS
+from .tournament import Tournament
+
+def replay(request):
+    if request.method == 'GET':
+        name = request.GET['name']
+        agent0 = request.GET['agent0']
+        agent1 = request.GET['agent1']
+        index = request.GET['index']
+        g = Game.objects.get(name=name, agent0=agent0, agent1=agent1, index=index)
+        json_data = g.replay
+        return HttpResponse(json_data)
+
+def query_game(request):
+    if request.method == 'GET':
+        filter_dict = {key: request.GET.get(key) for key in dict(request.GET).keys()}
+        result = Game.objects.filter(**filter_dict)
+        result = serializers.serialize('json', result, fields=('name', 'index', 'agent0', 'agent1', 'win', 'payoff'))
+        return HttpResponse(result)
+
+def query_payoff(request):
+    if request.method == 'GET':
+        filter_dict = {key: request.GET.get(key) for key in dict(request.GET).keys()}
+        result = Payoff.objects.filter(**filter_dict)
+        result = serializers.serialize('json', result)
+        return HttpResponse(result)
+
+
+@transaction.atomic
+def launch(request):
+    if request.method == 'GET':
+        eval_num = int(request.GET['eval_num'])
+        game = request.GET['name']
+        games_data, payoffs_data = Tournament(game, MODEL_IDS[game], eval_num).launch()
+        Game.objects.filter(name=game).delete()
+        Payoff.objects.filter(name=game).delete()
+        for game_data in games_data:
+            g = Game(name=game_data['name'],
+                     index=game_data['index'],
+                     agent0=game_data['agent0'],
+                     agent1=game_data['agent1'],
+                     win=game_data['win'],
+                     payoff=game_data['payoff'],
+                     replay=game_data['replay'])
+            g.save()
+        for payoff_data in payoffs_data:
+            p = Payoff(name=payoff_data['name'],
+                     agent0=payoff_data['agent0'],
+                     agent1=payoff_data['agent1'],
+                     payoff=payoff_data['payoff'])
+            p.save()
+        return HttpResponse(1)
+