tried video support (not working)

.gitignore

@@ -0,0 +1,6 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+video/

ql/__init__.py

@@ -8,7 +8,11 @@ import os
 from .settings import *
 
 env = gym.make(gym_name).env
+if use_video:
+    env = gym.wrappers.Monitor(env, './video/',video_callable=lambda episode_id: True,force=True)
 
 from .train import *
 from .play import *
 from .review import *
+
+frames = []

ql/play.py

@@ -26,6 +26,8 @@ def play(player_episodes=1):
                     done = True
 
                 ftext = ""
+                if use_video:
+                    env.render(mode="ansi")
                 if use_ansi:
                     ftext = env.render(mode="ansi")
                 else:

ql/review.py

@@ -6,27 +6,32 @@ def review(frames):
     rew = 0
     cnt = 1
 
-    for i, frame in enumerate(frames):
+    try:
-        print (u"{}[2J{}[;H".format(chr(27), chr(27)))
+        for i, frame in enumerate(frames):
-        print(f"Session: {frame['session']}")
+            out = u"{}[2J{}[;H".format(chr(27), chr(27))
-        print(frame['frame'])
+            out += f"Session: {frame['session']}\n"
-        print(f"Timestep: {i + 1}")
+            out += frame['frame']
-        print(f"State: {frame['state']}")
+            out += f"Timestep: {i + 1}\n"
-        print(f"Action: {frame['action']}")
+            out += f"State: {frame['state']}\n"
-        print(f"Reward: {frame['reward']}")
+            out += f"Action: {frame['action']}\n"
+            out += f"Reward: {frame['reward']}\n"
 
-        sess = frame['session']
+            sess = frame['session']
-        if sess != prevSess:
+            if sess != prevSess:
-            if rew > 0:
+                if rew > 0:
-                sucs += "+"
+                    sucs += "+"
-                sleep(1)
+                    sleep(1)
-            elif rew < 0:
+                elif rew < 0:
-                sucs += "-"
+                    sucs += "-"
-            else:
+                elif prevSess >= 0:
-                sucs += "."
+                    sucs += "."
-            prevSess = frame['session']
+                prevSess = frame['session']
-            cnt += 1
+                cnt += 1
-        rew = frame['reward']
+            rew = frame['reward']
 
-        print(f"\nSuccesses: [{sucs}]")
+            out += f"\nSuccesses: [{sucs}]\n"
-        sleep(.1)
+            print(out)
+            sleep(.1)
+
+    except KeyboardInterrupt:
+        return()

ql/settings.py

@@ -15,3 +15,4 @@ max_iterations = 1000
 # Render settings
 
 use_ansi = True
+use_video = False

ql/train.py

@@ -1,6 +1,9 @@
 from . import *
 
 def train(training_episodes=10000, resume=True):
+    global frames
+    frames = []
+
     if resume and os.path.exists(gym_name+".dat"):
         q_table = pickle.load( open( gym_name+".dat", "rb" ))
     else:
@@ -9,10 +12,12 @@ def train(training_episodes=10000, resume=True):
     episodes = training_episodes
     percentage = episodes / 100
 
-    frames = []
     suc_cnt = 0
 
     try:
+        print (u"{}[2J{}[;H".format(chr(27), chr(27)))
+        print("Training: 0%")
+        print("Successes so far: 0")
         for i in range(1, episodes + 1):
             state = env.reset()
 
@@ -41,6 +46,8 @@ def train(training_episodes=10000, resume=True):
 
                 if i % percentage == 0:
                     ftext = ""
+                    if use_video:
+                        env.render(mode="ansi")
                     if use_ansi:
                         ftext = env.render(mode="ansi")
                     else:
@@ -73,5 +80,3 @@ def train(training_episodes=10000, resume=True):
         pickle.dump(q_table , open( gym_name+".dat", "wb" ) )
         print(f"Q-table saved: {gym_name}_stopped.dat")
         exit()
-
-    return frames