def master():
db_conn = ibbdlib.get_db_conn(**db_server['223.4.155.172'])
db_conn.execute("TRUNCATE TABLE ibbd2.lib_cat_industry_keyword_2")
db_conn.close()
ibbdlib.migratetable.migrate_table(db_server['223.4.155.152'],
db_server['223.4.155.172'],
'ibbd2.user2')
ibbdlib.migratetable.migrate_table(db_server['223.4.155.152'],
db_server['223.4.155.172'],
'ibbd2.user_keywords')
ibbdlib.migratetable.migrate_table(db_server['223.4.155.152'],
db_server['223.4.155.172'],
'ibbd2.lib_cat_industry_keyword_2')
ibbdlib.migratetable.migrate_table_by_query(
db_server['223.4.155.152'], db_server['223.4.155.172'],
'ibbd2.user_shop_monitor_items', """SELECT user, shop_type,
(CASE WHEN shop_id='' THEN 0 ELSE shop_id END) shop_id,
user_num_id, item_id, status,
population_tsmp
FROM ibbd2.user_shop_monitor_items""")
# keep connection
dbConnPool = None
reConn = None
while True:
try:
dbConnPool = ConnectionPool(max_connections=1000,
host='localhost',
user='******',
passwd='ibbd_etl_secure',
db='topspider',
charset='utf8')
# reConn = getRedisConn2()
reConn = ibbdlib.get_redis_conn(**redis_server)
break
except Exception as e:
print e
runKeywordsSchedule(reConn, dbConnPool)
_ensure_schedule_complete(reConn)
dbConnPool.disconnect()
dbConnPool = ConnectionPool(max_connections=1000,
host='localhost',
user='******',
passwd='ibbd_etl_secure',
db='topspider',
charset='utf8')
while True:
try:
# migrateScheduleToRedis(reConn, dbConnPool)
break
except:
traceback.print_exc()
dbConnPool.disconnect()
del reConn
def main():
connpool = ConnectionPool(max_connections=20, **{
'host': 'localhost',
'user': '******',
'passwd': '',
'db': 'ibbd2',
})
reConn = getRedisConn2()
test_schedule_cat(reConn, connpool)
del reConn
connpool.disconnect()
def _reload_slave_db_pool(schedule_name):
global RE_CONN, DB_POOLS
host = None
while 1:
try:
db_server = json.loads(RE_CONN.get('slave_db_server'))
db_server = db_server.get(schedule_name) or \
db_server.get('default') or \
dict((k, v) for k, v in db_server.items() if not isinstance(v, dict))
host = db_server.get('host')
if not DB_POOLS.get(db_server.get('host')):
log.info('add slave db server %s' % db_server.get('host'))
DB_POOLS[db_server.get('host')] = ConnectionPool(
max_connections=1000, **db_server)
break
except:
traceback.print_exc()
time.sleep(10)
return DB_POOLS.get(host)
def prepar():
# get slave ID
global SLAVEID
SLAVEID = getMacId()
# get slave ID
global IP
IP = getMyIpInfo()['data']['ip']
global log
log_output = (slave_config.get(IP)['slave_info_log']
if slave_config.get(IP) else
slave_config.get('default')['slave_info_log'])
log = ibbdlib.ibbdlog.get_logger(
log_path='../log/',
log_file='slave %s %s.log' %
(PID, time.strftime('%Y%m%d', time.localtime())),
log_name='spider.slave',
msg_format=
'%(asctime)s\t%(name)s\t%(process)d\t%(levelname)s\t%(message)s',
output=log_output)
_slave_info('initing...')
# connection pool
conn_config = slave_config.get(IP) or slave_config.get('default')
conn_config = slave_config.get('default')
dbConnPool = ConnectionPool(
max_connections=1000,
**dict((k, v) for k, v in conn_config.get('mysql').items()
if not isinstance(v, dict)))
reConn = getRedisConn2(conn_config.get('redis'))
_slave_info('init... complete')
_slave_info('slave start to work')
global DB_POOLS, RE_CONN
DB_POOLS[conn_config.get('mysql')['host']] = dbConnPool
RE_CONN = reConn
def main():
_slave_info('initing...')
# get slave ID
global SLAVEID
SLAVEID = getMacId()
# get slave ID
global IP
IP = getMyIpInfo()['data']['ip']
# connection pool
conn_config = slave_config.get('IP') or slave_config.get('default')
conn_config = slave_config.get('default')
dbConnPool = ConnectionPool(max_connections=1000,
**conn_config.get('mysql'))
reConn = getRedisConn2(conn_config.get('redis'))
_slave_info('init... complete')
_slave_info('slave start to work')
dbConnPool.disconnect()
del reConn
@desc library
'''
import re
import time
import json
import uuid
import requests
import MySQLdb
import redis
from config import spider_header, mysql_db_config, redis_config, cookies
from dfhspider.connectionpool import ConnectionPool
SLAVEID = uuid.getnode()
connpool = ConnectionPool(max_connections=20, **mysql_db_config)
def getMyIpInfo():
refer = 'http://ip.taobao.com/service/getIpInfo.php'
try:
r = requests.request('POST', refer, data={'ip': 'myip'})
j = json.loads(r.text, encoding=r.encoding)
except Exception, e:
j = {'error': e.message}
return j
try:
IP = (getMyIpInfo().get('data') or {}).get('ip') or ''
except:
import ibbdlib
from dfhspider.app import getItemsByKeyword
from dfhspider.connectionpool import ConnectionPool
from dfhspider.topbll import saveKeywordSearchResult
db_server = {
'host': 'localhost',
'user': '******',
'passwd': 'ibbd_etl_secure',
'db': 'ibbd2',
'charset': 'utf8'
}
now = datetime.now()
db_pool = ConnectionPool(**db_server)
db_conn_test = db_pool.get_connection()
db_conn_test.close()
db_conn = ibbdlib.get_db_conn(**db_server)
query = """SELECT DISTINCT T2.keyword
FROM(
SELECT DISTINCT(keyword) AS keyword
FROM ibbd2.user_keywords T3
WHERE T3.status='1'
ORDER BY keyword
) T2
LEFT JOIN(
SELECT DISTINCT(keyword) AS keyword
FROM topspider.top_itemsearchresult T1
WHERE T1.population_tsmp > curdate()
def main():
# get log
global log
log = _get_log()
# slave info
_slave_info('initing...')
# get slave ID
global SLAVEID
SLAVEID = getMacId()
# get slave ID
global IP
IP = getMyIpInfo()['data']['ip']
# connection pool
conn_config = slave_config.get('IP') or slave_config.get('default')
# dbConnPool = ConnectionPool(max_connections=1000, **conn_config.get('mysql'))
dbConnPoolList = dict()
for (db_server_ip, db_server_config) in db_server.items():
dbConnPoolList[db_server_ip] = ConnectionPool(max_connections=1000, **db_server_config)
dbConnPoolList['master'] = ConnectionPool(max_connections=1000, **db_server_master)
reConn = getRedisConn2(conn_config.get('redis'))
_slave_info('init... complete')
_slave_info('slave start to work')
# thread pool
threadPool = []
while 1:
while 1:
try:
_slave_activity(reConn)
_slave_interval(reConn, 10, 10)
break
except Exception, e:
pass
for priority in sorted(SCHEDULE_PRIORITY.keys()):
# clear inactive Thread
threadPool = [childThread for childThread in threadPool if childThread.isAlive()]
for schedule in SCHEDULE_PRIORITY[priority]:
log.debug('schedule %s', schedule)
# how schedule should run
# 1.this schedule exists
# 2.schedule not active
# 3.current thread
# final:call schedule function by name
if reConn.exists(schedule) and schedule not in [t.name for t in threadPool] and sum(map(lambda t: t \
in [otherSchedule for otherSchedule in SCHEDULE_PRIORITY[priority] if otherSchedule
!= schedule], [thread.name for thread in threadPool])) == len(threadPool):
threadPool.append(threading.Thread(target=globals()[THREAD_TARGET[schedule]], name=schedule,
args=(reConn, dbConnPoolList)))
threadPool[-1].start()
time.sleep(10)
log.debug('thread waiting for %d', 30)