def flush_row(db, bucket, items):
try:
col = db[bucket]
for k, lines in items.iteritems():
parsed_lines = map(lambda line: json.loads(line), lines)
col.insert_many(parsed_lines)
except:
logger.error(full_stack())
return True
def flush_file(output_folder, bucket, items):
try:
bucket_folder = os.path.abspath('%s/%s' % (output_folder, bucket))
for k, lines in items.iteritems():
filename = os.path.abspath('%s/%s' % (bucket_folder, k))
with open(filename, 'ab+') as f:
for line in lines:
f.write('%s\n' % line)
logger.debug("flushed %d lines to %s" % (len(lines), filename))
except:
logger.error(full_stack())
return True
def redistribute_crawler_queue(self, crawler_id):
if (crawler_id in self.crawlers):
logger.warn('%s just failed... redistributing its workload'%(crawler_id))
try:
self.node_coordinator.distribute_to_nodes(self.crawlers[crawler_id]['crawler_queue'])
wait_timer = 180
# wait until it dies (flushed all the data...)
while(self.crawlers[crawler_id]['crawler'].is_alive() and wait_timer > 0):
time.sleep(60)
wait_timer -= 60
self.crawlers[crawler_id]['retry_timer_start_ts'] = int(time.time())
except Exception as exc:
logger.error(full_stack())
else:
logger.warn("whatever are you trying to do? crawler_id: [%s] is not valid..."%(crawler_id))
def flush_file(output_folder, bucket, items):
try:
bucket_folder = os.path.abspath("%s/%s" % (output_folder, bucket))
for k, lines in items.iteritems():
filename = os.path.abspath("%s/%s" % (bucket_folder, k))
with open(filename, "ab+") as f:
for line in lines:
f.write("%s\n" % line)
logger.debug("flushed %d lines to %s" % (len(lines), filename))
except:
logger.error(full_stack())
return True
parser = argparse.ArgumentParser()
parser.add_argument(
'-c',
'--config',
help=
"config.json that contains a) twitter api keys; b) redis connection string;",
required=True)
parser.add_argument('-p', '--proxies', help="the proxies.json file")
args = parser.parse_args()
proxies = None
if args.proxies:
with open(os.path.abspath(args.proxies), 'r') as proxy_f:
proxies = json.load(proxy_f)['proxies']
with open(os.path.abspath(args.config), 'r') as config_f:
config = json.load(config_f)
try:
start_server(config, proxies)
except KeyboardInterrupt:
print()
logger.error('You pressed Ctrl+C!')
pass
except Exception as exc:
logger.error(exc)
logger.error(full_stack())
finally:
pass
if cmd:
scheduler.enqueue(cmd)
if __name__=="__main__":
parser = argparse.ArgumentParser()
parser.add_argument('-c', '--config', help="config.json that contains a) twitter api keys; b) redis connection string;", required = True)
parser.add_argument('-p', '--proxies', help="the proxies.json file")
args = parser.parse_args()
proxies = None
if args.proxies:
with open(os.path.abspath(args.proxies), 'rb') as proxy_f:
proxies = json.load(proxy_f)['proxies']
with open(os.path.abspath(args.config), 'rb') as config_f:
config = json.load(config_f)
try:
start_server(config, proxies)
except KeyboardInterrupt:
print()
logger.error('You pressed Ctrl+C!')
pass
except Exception as exc:
logger.error(exc)
logger.error(full_stack())
finally:
pass
