import itertools
import random
import sqlite3
import common
nusers = 1
total_votes_per_user = common.schedule_ndays() * common.max_votes_per_day
input_connection, input_cursor, output_connection, output_cursor = common.io_connections_and_cursors()
vote_id = common.next_output_id(output_cursor)
nanswers = next(input_cursor.execute(
'SELECT COUNT(*) FROM posts WHERE PostTypeId = ?',
(common.Posts_PostTypeId_Answer,)))[0]
# Second puppet only.
for user_row in itertools.islice(common.iterate_users(), 1, 1 + nusers):
user_id = user_row[common.users_csv_user_id_col]
nscheduled_answers = 0
answer_indexes = set()
while nscheduled_answers < total_votes_per_user:
n = random.randint(0, nanswers - 1)
if n not in answer_indexes:
answer_indexes.add(n)
nscheduled_answers += 1
user_nvotes = 0
for answer_i in answer_indexes:
post_row = next(input_cursor.execute(
'SELECT * FROM posts WHERE PostTypeId = ? LIMIT 1 OFFSET ?',
(common.Posts_PostTypeId_Answer, answer_i)))
output_cursor.execute('INSERT INTO votes VALUES (?, ?, ?, ?, NULL, NULL)',
(vote_id, user_id, post_row['Id'], post_row['ParentId']))
first_user_range = 1
# Last user to schedule for, exclusive.
after_last_user_range = 2
total_votes_per_user = common.schedule_ndays() * common.max_votes_per_day
input_connection, input_cursor, output_connection, output_cursor = common.io_connections_and_cursors()
vote_id = common.next_output_id(output_cursor)
post_rows = input_cursor.execute("""
SELECT answers.* FROM posts AS answers
INNER JOIN posts AS questions
WHERE questions.PostTypeId = ?
AND answers.PostTypeId = ?
AND answers.ParentId = questions.Id
AND answers.CommunityOwnedDate IS NULL
AND questions.Tags LIKE '%<git>%'
ORDER BY Id ASC
LIMIT ?;
""",
(common.Posts_PostTypeId_Question, common.Posts_PostTypeId_Answer, total_votes_per_user))
for user_row in itertools.islice(common.iterate_users(), first_user_range, after_last_user_range):
user_id = user_row[common.users_csv_user_id_col]
user_nvotes = 0
for post_row in post_rows:
output_cursor.execute('INSERT INTO votes VALUES (?, ?, ?, ?, NULL, NULL)',
(vote_id, user_id, post_row['Id'], post_row['ParentId']))
vote_id += 1
user_nvotes += 1
if user_nvotes != total_votes_per_user:
print('Warning: not enough posts.')
common.commit_and_closeclose_io_connections(input_connection, output_connection)
input_connection, input_cursor, output_connection, output_cursor = common.io_connections_and_cursors(
)
vote_id = common.next_output_id(output_cursor)
post_rows = input_cursor.execute(
"""
SELECT answers.* FROM posts AS answers
INNER JOIN posts AS questions
WHERE questions.PostTypeId = ?
AND answers.PostTypeId = ?
AND answers.ParentId = questions.Id
AND answers.CommunityOwnedDate IS NULL
AND questions.Tags LIKE '%<git>%'
ORDER BY Id ASC
LIMIT ?;
""", (common.Posts_PostTypeId_Question, common.Posts_PostTypeId_Answer,
total_votes_per_user))
for user_row in itertools.islice(common.iterate_users(), first_user_range,
after_last_user_range):
user_id = user_row[common.users_csv_user_id_col]
user_nvotes = 0
for post_row in post_rows:
output_cursor.execute(
'INSERT INTO votes VALUES (?, ?, ?, ?, NULL, NULL)',
(vote_id, user_id, post_row['Id'], post_row['ParentId']))
vote_id += 1
user_nvotes += 1
if user_nvotes != total_votes_per_user:
print('Warning: not enough posts.')
common.commit_and_closeclose_io_connections(input_connection,
output_connection)
connection.close()
# Is thread safe: http://stackoverflow.com/questions/2973900/is-pythons-logging-module-thread-safe
logging.basicConfig(
filename = os.path.splitext(os.path.realpath(__file__))[0] + '.log',
level = logging.DEBUG,
format = '%(asctime)s | %(levelname)-7s | %(threadName)7s | %(message)s',
)
logging.Formatter.converter = time.gmtime
script_dir = os.path.dirname(os.path.realpath(__file__))
logging.debug(
'Last git commit SHA = ' +
subprocess.check_output(['git', '-C', script_dir, 'rev-parse', 'HEAD']))
torrc_dir = os.path.join(script_dir, 'torrc')
if len(sys.argv) > 1:
casperjs_path = sys.argv[1]
else:
casperjs_path = '/home/ciro/.nvm/v0.10.26/bin/casperjs'
with open(common.users_csv_path, 'r') as user_file:
user_csv = csv.reader(user_file)
threads = []
for i, user_row in enumerate(common.iterate_users()):
t = UserVotesThread(user_row, i)
threads.append(t)
t.start()
for t in threads:
t.join()
# Is thread safe: http://stackoverflow.com/questions/2973900/is-pythons-logging-module-thread-safe
logging.basicConfig(
filename=os.path.splitext(os.path.realpath(__file__))[0] + '.log',
level=logging.DEBUG,
format='%(asctime)s | %(levelname)-7s | %(threadName)7s | %(message)s',
)
logging.Formatter.converter = time.gmtime
script_dir = os.path.dirname(os.path.realpath(__file__))
logging.debug(
'Last git commit SHA = ' +
subprocess.check_output(['git', '-C', script_dir, 'rev-parse', 'HEAD']))
torrc_dir = os.path.join(script_dir, 'torrc')
if len(sys.argv) > 1:
casperjs_path = sys.argv[1]
else:
casperjs_path = '/home/ciro/.nvm/v0.10.26/bin/casperjs'
with open(common.users_csv_path, 'r') as user_file:
user_csv = csv.reader(user_file)
threads = []
for i, user_row in enumerate(common.iterate_users()):
t = UserVotesThread(user_row, i)
threads.append(t)
t.start()
for t in threads:
t.join()