Repository navigation
Expand file tree
/
Copy pathprepare_bigquery.py
More file actions
234 lines (194 loc) · 7.8 KB
/
Copy pathprepare_bigquery.py
File metadata and controls
234 lines (194 loc) · 7.8 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
# Copyright 2018 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Script to process debug SGFs for upload to BigQuery.
Handles one model per invocation, for easy sharding of work.
Usage:
python oneoffs/prepare_bigquery.py 000001-model-name
The load commands look like:
bq load --project_id=$PROJECT_ID \
--source_format=NEWLINE_DELIMITED_JSON \
$PROJECT_ID:minigo_v5_19.games \
gs://$BUCKET_NAME/bigquery/holdout/games/* \
oneoffs/bigquery_games_schema.json
bq load --project_id=$PROJECT_ID \
--source_format=NEWLINE_DELIMITED_JSON \
$PROJECT_ID:minigo_v5_19.moves \
gs://$BUCKET_NAME/bigquery/holdout/moves/* \
oneoffs/bigquery_moves_schema.json
"""
import sys
sys.path.insert(0, '.')
import collections
import json
import os
import re
from absl import app, flags
from tensorflow import gfile
from tqdm import tqdm
import coords
import sgf_wrapper
from rl_loop import shipname
import utils
DebugRow = collections.namedtuple('DebugRow', [
'move', 'action', 'Q', 'U', 'prior', 'orig_prior', 'N', 'soft_N', 'p_delta', 'p_rel'])
BUCKET_NAME = os.environ['BUCKET_NAME']
BASE_GS_DIR = 'gs://{}'.format(BUCKET_NAME)
HOLDOUT_PATH = os.path.join('{}', 'data', 'holdout', '{}')
PATH_TEMPLATE = os.path.join('{}', 'sgf', '{}', 'full', '{}')
OUTPUT_PATH = os.path.join('{}', 'bigquery', 'holdout', '{}', '{}')
flags.DEFINE_string("base_dir", BASE_GS_DIR, "base directory for minigo data")
flags.DEFINE_boolean("only_top_move", False,
"only include policy and playout data about played move")
FLAGS = flags.FLAGS
def get_sgf_names(model):
game_dir = HOLDOUT_PATH.format(FLAGS.base_dir, model)
tf_records = map(os.path.basename, gfile.ListDirectory(game_dir))
sgfs = [record.replace('.tfrecord.zz', '.sgf') for record in tf_records]
return [PATH_TEMPLATE.format(FLAGS.base_dir, model, sgf) for sgf in sgfs]
def extract_holdout_model(model):
game_output_path = OUTPUT_PATH.format(FLAGS.base_dir, 'games', model)
move_output_path = OUTPUT_PATH.format(FLAGS.base_dir, 'moves', model)
gfile.MakeDirs(os.path.basename(game_output_path))
gfile.MakeDirs(os.path.basename(move_output_path))
with gfile.GFile(game_output_path, 'w') as game_f, \
gfile.GFile(move_output_path, 'w') as move_f:
for sgf_name in tqdm(get_sgf_names(model)):
game_data, move_data = extract_data(sgf_name)
game_f.write(json.dumps(game_data) + '\n')
for move_datum in move_data:
move_f.write(json.dumps(move_datum) + '\n')
def extract_data(filename):
with gfile.GFile(filename) as f:
contents = f.read()
root_node = sgf_wrapper.get_sgf_root_node(contents)
game_data = extract_game_data(filename, root_node)
move_data = extract_move_data(
root_node,
game_data['worker_id'],
game_data['completed_time'],
game_data['board_size'])
return game_data, move_data
def extract_game_data(gcs_path, root_node):
props = root_node.properties
komi = float(sgf_wrapper.sgf_prop(props.get('KM')))
result = sgf_wrapper.sgf_prop(props.get('RE', ''))
board_size = int(sgf_wrapper.sgf_prop(props.get('SZ')))
value = utils.parse_game_result(result)
was_resign = '+R' in result
filename = os.path.basename(gcs_path)
filename_no_ext, _ = os.path.splitext(filename)
# BigQuery's TIMESTAMP() takes in unix millis.
completion_millis = 1000 * int(filename_no_ext.split('-')[0])
worker_id = filename_no_ext.split('-')[-1]
model_num = shipname.detect_model_num(props.get('PW')[0])
sgf_url = gcs_path
first_comment_node_lines = root_node.next.properties['C'][0].split('\n')
# in-place edit to comment node so that first move's comment looks
# the same as all the other moves.
root_node.next.properties['C'][0] = '\n'.join(first_comment_node_lines[1:])
resign_threshold = float(first_comment_node_lines[0].split()[-1])
return {
'worker_id': worker_id,
'completed_time': completion_millis,
'board_size': board_size,
'model_num': model_num,
'result_str': result,
'value': value,
'was_resign': was_resign,
'sgf_url': sgf_url,
'resign_threshold': resign_threshold,
}
def extract_move_data(root_node, worker_id, completed_time, board_size):
current_node = root_node.next
move_data = []
move_num = 1
while current_node is not None:
props = current_node.properties
if 'B' in props:
to_play = 1
move_played = props['B'][0]
elif 'W' in props:
to_play = -1
move_played = props['W'][0]
else:
import pdb; pdb.set_trace()
move_played = coords.to_flat(coords.from_sgf(move_played))
post_Q, debug_rows = parse_comment_node(props['C'][0])
def get_row_data(debug_row):
column_names = ["prior", "orig_prior", "N", "soft_N"]
return [getattr(debug_row, field) for field in column_names]
if FLAGS.only_top_move:
assert len(debug_rows) <= 1
row_data = list(map(get_row_data, debug_rows))
else:
row_data = [[0] * 4 for _ in range(board_size * board_size + 1)]
for debug_row in debug_rows:
move = debug_row.move
row_data[move] = get_row_data(debug_row)
policy_prior, policy_prior_orig, mcts_visits, mcts_visits_norm = \
zip(*row_data)
move_data.append({
'worker_id': worker_id,
'completed_time': completed_time,
'move_num': move_num,
'turn_to_play': to_play,
'move': move_played,
'move_kgs': coords.to_gtp(coords.from_flat(move_played)),
'prior_Q': None,
'post_Q': post_Q,
'policy_prior': policy_prior,
'policy_prior_orig': policy_prior_orig,
'mcts_visit_counts': mcts_visits,
'mcts_visit_counts_norm': mcts_visits_norm,
})
move_num += 1
current_node = current_node.next
return move_data
def parse_comment_node(comment):
# Example of a comment node. The resign threshold line appears only
# for the first move in the game; it gets preprocessed by extract_game_data
"""
Resign Threshold: -0.88
-0.0662
D4 (100) ==> D16 (14) ==> Q16 (3) ==> Q4 (1) ==> Q: -0.07149
move: action Q U P P-Dir N soft-N p-delta p-rel
D4 : -0.028, -0.048, 0.020, 0.048, 0.064, 100 0.1096 0.06127 1.27
D16 : -0.024, -0.043, 0.019, 0.044, 0.059, 96 0.1053 0.06135 1.40
"""
lines = comment.split('\n')
if lines[0].startswith('Resign'):
lines = lines[1:]
post_Q = float(lines[0])
debug_rows = []
comment_splitter = re.compile(r'[ :,]')
for line in lines[3:]:
if not line:
continue
columns = comment_splitter.split(line)
columns = list(filter(bool, columns))
coord, *other_columns = columns
coord = coords.to_flat(coords.from_gtp(coord))
debug_rows.append(DebugRow(coord, *map(float, other_columns)))
if FLAGS.only_top_move:
break
return post_Q, debug_rows
def main(unusedargv):
if len(unusedargv) != 2:
print("Usage: python {} 000001-model-name")
sys.exit(1)
model = unusedargv[1]
extract_holdout_model(model)
if __name__ == '__main__':
app.run(main)