def watchlist(self):
'''Watch a fixed list of user_id'''
# Clean up database first
db = misc.mysql_db(self.config['db_server'],
self.config['db_username'],
self.config['db_password'],
self.config['db_database'], self.logger)
stmt = 'DELETE FROM target_users'
db.execute(stmt)
# TODO still have problem
stmt = ('LOAD DATA LOCAL INFILE "seed.lst" INTO TABLE target_users '
'FIELDS TERMINATED BY \"\\t\" LINES TERMINATED BY \"\\n\"')
db.execute(stmt)
# Get that list first
self.crawl('seed.lst')
# Get that list's friend second
stmt = ('SELECT DISTINCT friend_id FROM friends, target_users '
'WHERE friends.user_id = target_users.user_id')
db.execute(stmt)
results = db.cursor.fetchall()
db.__del__()
misc.write_to_files(results, 'initial_friends',
self.config['seed_per_file'], 'utf')
# Enter the generate-crawl-update loop
self.twalerloop()
def generate(self):
try:
# TODO improve our naive seed generation method
#---------------------------------
# select the top _seed_limit_ most re-occuring friend that
# has yet to be crawled
stmt = ("SELECT friend_id FROM friends WHERE friend_id NOT IN "
"(SELECT user_id FROM users_update) "
"group by friend_id order by count(*) desc limit %s" %
self.config['seed_limit'])
self.db.execute(stmt)
self.logger.debug("MySQL generate_users Query Complete")
seedType = 'utf'
#---------------------------------
# fetch results and write to new seed file
results = self.db.cursor.fetchall()
timestamp = misc.timefunctions.datestamp()
misc.write_to_files(results, 'seeds_' + timestamp,
self.config['seed_per_file'], seedType)
except Exception as e:
traceback.print_stack()
self.logger.error(str(e))
