def _render(self, mode='human', close=False):
if close:
return
outfile = StringIO() if mode == 'ansi' else sys.stdout
out = self.desc.copy().tolist()
out = [[c.decode('utf-8') for c in line] for line in out]
taxirow, taxicol, passidx, destidx = self.decode(self.s)
def ul(x):
return "_" if x == " " else x
if passidx < 4:
out[1 + taxirow][2 * taxicol + 1] = utils.colorize(out[1 + taxirow][2 * taxicol + 1], 'yellow',
highlight=True)
pi, pj = self.locs[passidx]
out[1 + pi][2 * pj + 1] = utils.colorize(out[1 + pi][2 * pj + 1], 'blue', bold=True)
else: # passenger in taxi
out[1 + taxirow][2 * taxicol + 1] = utils.colorize(ul(out[1 + taxirow][2 * taxicol + 1]), 'green',
highlight=True)
di, dj = self.locs[destidx]
out[1 + di][2 * dj + 1] = utils.colorize(out[1 + di][2 * dj + 1], 'magenta')
outfile.write("\n".join(["".join(row) for row in out]) + "\n")
if self.lastaction is not None:
outfile.write(" ({})\n".format(["South", "North", "East", "West", "Pickup", "Dropoff"][self.lastaction]))
else:
outfile.write("\n")
# No need to return anything for human
if mode != 'human':
return outfile
def _render_observation(self):
x = self.read_head_position
x_str = "Observation Tape : "
for i in range(-2, self.input_width + 2):
if i == x:
x_str += colorize(self._get_str_obs(np.array([i])), 'green', highlight=True)
else:
x_str += self._get_str_obs(np.array([i]))
x_str += "\n"
return x_str
def _render_observation(self):
x = self.read_head_position
label = "Observation Grid : "
x_str = ""
for j in range(-1, self.rows + 1):
if j != -1:
x_str += " " * len(label)
for i in range(-2, self.input_width + 2):
if i == x[0] and j == x[1]:
x_str += colorize(self._get_str_obs((i, j)), 'green', highlight=True)
else:
x_str += self._get_str_obs((i, j))
x_str += "\n"
x_str = label + x_str
return x_str
def _render(self, mode='human', close=False):
if close:
return
outfile = StringIO() if mode == 'ansi' else sys.stdout
row, col = self.s // self.ncol, self.s % self.ncol
desc = self.desc.tolist()
desc = [[c.decode('utf-8') for c in line] for line in desc]
desc[row][col] = utils.colorize(desc[row][col], "red", highlight=True)
if self.lastaction is not None:
outfile.write(" ({})\n".format(["Left","Down","Right","Up"][self.lastaction]))
else:
outfile.write("\n")
outfile.write("\n".join(''.join(line) for line in desc)+"\n")
return outfile
def _render(self, mode='human', close=False):
if close:
# Nothing interesting to close
return
outfile = StringIO() if mode == 'ansi' else sys.stdout
inp = "Total length of input instance: %d, step: %d\n" % (self.input_width, self.time)
outfile.write(inp)
x, y, action = self.read_head_position, self.write_head_position, self.last_action
if action is not None:
inp_act, out_act, pred = action
outfile.write("=" * (len(inp) - 1) + "\n")
y_str = "Output Tape : "
target_str = "Targets : "
if action is not None:
pred_str = self.charmap[pred]
x_str = self._render_observation()
for i in range(-2, len(self.target) + 2):
target_str += self._get_str_target(i)
if i < y - 1:
y_str += self._get_str_target(i)
elif i == (y - 1):
if action is not None and out_act == 1:
color = 'green' if pred == self.target[i] else 'red'
y_str += colorize(pred_str, color, highlight=True)
else:
y_str += self._get_str_target(i)
outfile.write(x_str)
outfile.write(y_str + "\n")
outfile.write(target_str + "\n\n")
if action is not None:
outfile.write("Current reward : %.3f\n" % self.last_reward)
outfile.write("Cumulative reward : %.3f\n" % self.episode_total_reward)
move = self.MOVEMENTS[inp_act]
outfile.write("Action : Tuple(move over input: %s,\n" % move)
out_act = out_act == 1
outfile.write(" write to the output tape: %s,\n" % out_act)
outfile.write(" prediction: %s)\n" % pred_str)
else:
outfile.write("\n" * 5)
return outfile