def sample_positions_from_games(sgf_files, num_positions=1):
pos_data = []
move_data = []
result_data = []
move_idxs = []
fail_count = 0
for path in tqdm(sgf_files, desc="loading sgfs", unit="games"):
try:
positions, moves, results = oneoff_utils.parse_sgf(path)
except KeyboardInterrupt:
raise
except Exception as e:
print("Parse exception:", e)
fail_count += 1
continue
# add entire game
if num_positions == -1:
pos_data.extend(positions)
move_data.extend(moves)
move_idxs.extend(range(len(positions)))
result_data.extend(results)
else:
for idx in np.random.choice(len(positions), num_positions):
pos_data.append(positions[idx])
move_data.append(moves[idx])
result_data.append(results[idx])
move_idxs.append(idx)
print("Sampled {} positions, failed to parse {} files".format(
len(pos_data), fail_count))
return pos_data, move_data, result_data, move_idxs
def subsample():
"""Sample num_positions postions from each game in sgf_dir
Usage:
python3 sharp_positions.py subsample --num_positions 10 --sgf_dir data/s
NOTE(sethtroisi): see link for a script to truncate SGFs at move number
https://github.com/sethtroisi/go-scripts
"""
sgf_files = oneoff_utils.find_and_filter_sgf_files(FLAGS.sgf_dir, None,
None)
with open(FLAG.collection, 'w') as collection:
fails = 0
for path in tqdm(sorted(sgf_files)):
try:
positions, moves, results = oneoff_utils.parse_sgf(path)
except KeyboardInterrupt:
raise
except Exception as e:
fails += 1
print("Fail {}, while parsing {}: {}".format(fails, path, e))
continue
moves = len(positions)
indexes = random.sample(range(10, moves), FLAGS.num_positions)
for index in sorted(indexes):
collection.write('{}, {}\n'.format(path, index))
def positions_from_sgfs(sgf_files):
# sgf_replay doesn't like empty sgf file
data = [("empty", go.Position(komi=7.5))]
for sgf in sgf_files:
sgf_name = os.path.basename(sgf).replace(".sgf", "")
positions, moves, _ = oneoff_utils.parse_sgf(sgf)
final = positions[-1].play_move(moves[-1])
data.append((sgf_name, final))
return data
def positions_from_sgfs(sgf_files, include_empty_position=True):
data = []
if include_empty:
# sgf_replay doesn't like SGFs with no moves played.
# Add the empty position for analysis manually.
data.append(("empty", go.Position(komi=7.5)))
for sgf in sgf_files:
sgf_name = os.path.basename(sgf).replace(".sgf", "")
positions, moves, _ = oneoff_utils.parse_sgf(sgf)
final = positions[-1].play_move(moves[-1])
data.append((sgf_name, final))
return data
