def file_write_redis():
'''
将源文件中的内容写入redis中
:return:
'''
gaode_file_path = file_exists() #
# fanyule_two_game_file_path = r'/ftp_samba/112/file_4spider/dmn_fanyule2_game/' #
startup_nodes = [{'host': 'redis1', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
# 高德
gaode_file = open(gaode_file_path, 'r')
gaode_keyword_length = r.llen(
'spider:python:gaode:keyword') # redis中gaode的数据量
print 'redis中gaode_keyword列表长度：', gaode_keyword_length
if gaode_keyword_length != 0:
r.delete('spider:python:gaode:keyword')
print 'redis中gaode_keyword列表长度不为0, 删除后的列表长度：', r.llen(
'spider:python:gaode:keyword')
for line in gaode_file:
new_line = line.strip()
if new_line:
r.rpush('spider:python:gaode:keyword', new_line)
gaode_keyword_length = r.llen('spider:python:gaode:keyword')
print '重新写入后redis中gaode_keyword列表长度：', gaode_keyword_length
def main():
lock = threading.Lock()
print time.strftime('[%Y-%m-%d %H:%M:%S]：'), 'start'
startup_nodes = [{'host': 'redis3', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
print 'redis中pubg_death昵称列表长度：', r.llen('spider:python:pubg_death:keyword')
get_redis_proxy() # 将redis中的代理ip放入到PROXY_IP_Q队列中
proxy_count = PROXY_IP_Q.qsize()
print time.strftime('[%Y-%m-%d %H:%M:%S]'), '代理ip队列中的ip数量：', proxy_count
print time.strftime('[%Y-%m-%d %H:%M:%S]'), '开启的线程数：', proxy_count
threads = []
for i in xrange(50):
t = threading.Thread(target=pubg_userid, args=(lock, r))
t.start()
threads.append(t)
for t in threads:
t.join()
print '源数据剩余量：', r.llen('spider:python:pubg_death:keyword')
print '抓取结束'
def main():
startup_nodes = [{'host': 'redis3', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
gaode_length = r.llen('spider:python:gaode:keyword:dest') # redis的数据量
print time.strftime(
'[%Y-%m-%d %H:%M:%S]'), 'redis中gaode的数据量：', gaode_length
lock = threading.Lock()
first_hour = time.strftime('%H')
date = time.strftime('%Y%m%d')
dest_path = '/ftp_samba/112/spider/python/gd_location/' # 线上数据存储文件
if not os.path.exists(dest_path):
os.makedirs(dest_path)
dest_file_name = os.path.join(dest_path, 'gd_location_' + date)
fileout = open(dest_file_name, 'a')
threads = []
for i in xrange(1):
t = threading.Thread(target=gaode, args=(lock, r, first_hour, fileout))
t.start()
threads.append(t)
for t in threads:
t.join()
try:
fileout.flush()
fileout.close()
except IOError as e:
time.sleep(2)
fileout.close()
def main():
startup_nodes = [{'host': 'redis3', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
pubg_friends_length = r.llen(
'spider:python:pubg_friends:keyword:dest') # redis的数据量
print time.strftime(
'[%Y-%m-%d %H:%M:%S]'), 'redis中pubg_friends的数据量：', pubg_friends_length
lock = threading.Lock()
first_hour = time.strftime('%H') # 小时
date = time.strftime('%Y%m%d') # 数据文件日期
dest_path = '/ftp_samba/112/spider/python/pubg' # linux上的文件目录
if not os.path.exists(dest_path):
os.makedirs(dest_path)
dest_file_name = os.path.join(dest_path, 'pubg_friends_' + date)
fileout = open(dest_file_name, 'a')
threads = []
for i in xrange(1):
t = threading.Thread(target=pubg_friends,
args=(lock, r, first_hour, fileout))
t.start()
threads.append(t)
for t in threads:
t.join()
try:
fileout.flush()
fileout.close()
except IOError as e:
time.sleep(2)
fileout.close()
print time.strftime('[%Y-%m-%d %H:%M:%S]：'), 'over'
def redis_cluster():
'''集群操作'''
redis_nodes = [{
'host': '10.12.28.222',
'port': 6380
}, {
'host': '10.12.28.222',
'port': 6381
}, {
'host': '10.12.28.224',
'port': 6380
}, {
'host': '10.12.28.224',
'port': 6381
}, {
'host': '10.12.28.227',
'port': 6380
}, {
'host': '10.12.28.227',
'port': 6381
}]
try:
r = StrictRedisCluster(startup_nodes=redis_nodes)
except Exception as e:
print("connect error %s" % e)
# string 操作
r.set('thoth:thoth-ai:robot:1', 'kk')
# r.delete('thoth:thoth-ai:robot:1')
print("name is", r.get('thoth:thoth-ai:robot:1'))
# list 操作
r.lpush('thoth:thoth-ai:robot:2', [[1, 2, 3], [2, 3, 4]])
print('list len:', r.llen("thoth:thoth-ai:robot:2")) # list size
print("list ", r.lindex('thoth:thoth-ai:robot:2', 0))
# hash 操作
r.hset('thoth:thoth-ai:robot:3', 'private_vector', [[1, 2, 3], [2, 3, 4]])
r.hset('thoth:thoth-ai:robot:3', 'public_vector', [['4', 3, 2], [0, 1, 1]])
pv = r.hget(
'thoth:thoth-ai:robot:3',
'public_vector',
)
print('hash.robot3.public_vector:', pv)
aaa = pv.decode('utf-8')
print(type(aaa), aaa)
b = eval(aaa) # eval 函数妙用：将string‘[1,2,3]’--->list [1,2,3]
print(type(b), b)
def exec_redis(redis_nodes,cmd_list):
'''
redis_nodes = [{'host':'10.101.104.132','port':1321},
{'host':'10.101.104.132','port':1322},
{'host':'10.101.104.132','port':1323}
{'host':'10.101.104.132','port':1323,'password':None,'db':0}
]
cmd_list = [ #暂时只支持read，而且只有这几种
"get key1",
"llen lis1",
"smembers set1",
"scard set1",
"hgetall hashtable",
"lindex list1 2",
"hget hashtable key1",
"lrange list1 1 3",
]
'''
redisconn = StrictRedisCluster(startup_nodes=redis_nodes,decode_responses=True)
res = {}
for cmd in cmd_list:
cmd = cmd.strip()
arr = re.split(" +",cmd)
type = arr[0].lower()
if type == 'get':
tmp = redisconn.get(arr[1])
elif type == 'hgetall':
tmp = redisconn.hgetall(arr[1])
elif type == 'llen':
tmp = redisconn.llen(arr[1])
elif type == 'smembers':
tmp = redisconn.smembers(arr[1])
elif type == 'scard':
tmp = redisconn.scard(arr[1])
elif type == 'hget':
tmp = redisconn.hget(arr[1],arr[2])
elif type == 'lrange':
tmp = redisconn.lrange(arr[1],arr[2],arr[3])
elif type == 'lindex':
tmp = redisconn.lrange(arr[1],arr[2])
else:
tmp = 'this cmd is not support yeild'
res[cmd] = [json_decode(tmp)]
return res
def get_redis_proxy():
'''
从redis相应的key中获取代理ip
:return:
'''
startup_nodes = [{'host': 'redis1', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
weibo_user_proxy_length = r.llen('spider:weibo_user:proxy') # weibo_user
print time.strftime('[%Y-%m-%d %H:%M:%S]'), 'redis中weibo_user的代理ip长度：', weibo_user_proxy_length
if weibo_user_proxy_length == 0:
print time.strftime('[%Y-%m-%d %H:%M:%S]'), 'redis中的代理ip数量为0，等待60s'
time.sleep(60)
return get_redis_proxy()
for i in xrange(weibo_user_proxy_length):
ip = r.lpop('spider:weibo_user:proxy')
proxies = {
'http': "http://*****:*****@{ip}".format(ip=ip),
'https': "http://*****:*****@{ip}".format(ip=ip)
}
PROXY_IP_Q.put(proxies)
def main():
lock = Lock()
print time.strftime('[%Y-%m-%d %H:%M:%S]'), 'start'
startup_nodes = [{'host': 'redis1', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
print 'redis中gaode_keyword来源列表长度：', r.llen('spider:python:gaode:keyword')
key_test() # 将有效的key放入KEYS_QUEUE
key = KEYS_QUEUE.get()
print '获取的第一个key：', key
threads = []
for i in xrange(50):
t = threading.Thread(target=gaode, args=(lock, key, r))
t.start()
threads.append(t)
for t in threads:
t.join()
print time.strftime('[%Y-%m-%d %H:%M:%S]'), '抓取结束'
def get_redis_proxy():
'''
从redis相应的key中获取代理ip(读取快代理的代理ip)
:return:
'''
startup_nodes = [{'host': 'redis2', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
baidu_zhishu_proxy_length = r.llen(
'spider:baidu_zhishu:proxy:kuai') # baidu_zhishu
print time.strftime(
'[%Y-%m-%d %H:%M:%S]'
), 'redis中baidu_zhishu的代理ip长度：', baidu_zhishu_proxy_length
if baidu_zhishu_proxy_length == 0:
print time.strftime('[%Y-%m-%d %H:%M:%S]'), 'redis中的代理ip数量为0，等待60s'
time.sleep(60)
return get_redis_proxy()
for i in xrange(20):
ip = r.lpop('spider:baidu_zhishu:proxy:kuai')
proxies = {
'http': "http://{ip}".format(ip=ip),
# 'https': "http://{ip}".format(ip=ip)
}
PROXY_IP_Q.put(proxies)
def file_write_redis():
'''
将源文件中的内容写入redis中
:return:
'''
# wrd_keyword_file_path = file_exists() # 数据源文件路径
pubg_nickname_file_path = r'/ftp_samba/112/file_4spider/pubg_nickname/pubg_nickname' # 数据源文件路径
startup_nodes = [{'host': 'redis2', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
pubg_nickname_file = open(pubg_nickname_file_path, 'r')
pubg_friends_nickname_length = r.llen('spider:python:pubg_friends:keyword')
print 'redis中pubg_friends_nickname列表长度：', pubg_friends_nickname_length
if pubg_friends_nickname_length < 200000:
for line in pubg_nickname_file:
r.rpush('spider:python:pubg_friends:keyword', line.strip())
pubg_friends_nickname_length = r.llen(
'spider:python:pubg_friends:keyword')
print '重新写入后redis中pubg_friends_nickname列表长度：', pubg_friends_nickname_length
pubg_nickname_file = open(pubg_nickname_file_path, 'r')
pubg_match_nickname_length = r.llen('spider:python:pubg_match:keyword')
print 'redis中pubg_match_nickname列表长度：', pubg_match_nickname_length
if pubg_match_nickname_length < 200000:
for line in pubg_nickname_file:
r.rpush('spider:python:pubg_match:keyword', line.strip())
pubg_match_nickname_length = r.llen('spider:python:pubg_match:keyword')
print '重新写入后redis中pubg_match_nickname列表长度：', pubg_match_nickname_length
pubg_nickname_file = open(pubg_nickname_file_path, 'r')
pubg_death_nickname_length = r.llen('spider:python:pubg_death:keyword')
print 'redis中pubg_death_nickname列表长度：', pubg_death_nickname_length
if pubg_death_nickname_length < 200000:
for line in pubg_nickname_file:
r.rpush('spider:python:pubg_death:keyword', line.strip())
pubg_death_nickname_length = r.llen('spider:python:pubg_death:keyword')
print '重新写入后redis中pubg_death_nickname列表长度：', pubg_death_nickname_length
class RedisDB():
def __init__(self, ip_ports = IP_PORTS, db = DB, user_pass = USER_PASS):
# super(RedisDB, self).__init__()
if not hasattr(self,'_redis'):
self._is_redis_cluster = False
try:
if len(ip_ports) > 1:
startup_nodes = []
for ip_port in ip_ports:
ip, port = ip_port.split(':')
startup_nodes.append({"host":ip, "port":port})
self._redis = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
self._pipe = self._redis.pipeline(transaction=False)
self._is_redis_cluster = True
else:
ip, port = ip_ports[0].split(':')
self._redis = redis.Redis(host = ip, port = port, db = db, password = user_pass, decode_responses=True) # redis默认端口是6379
self._pipe = self._redis.pipeline(transaction=True) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
except Exception as e:
raise
else:
log.info('连接到redis数据库 %s'%(tools.dumps_json(ip_ports)))
def sadd(self, table, values):
'''
@summary: 使用无序set集合存储数据， 去重
---------
@param table:
@param values: 值； 支持list 或 单个值
---------
@result: 若库中存在 返回0，否则入库，返回1。 批量添加返回None
'''
if isinstance(values, list):
if not self._is_redis_cluster: self._pipe.multi()
for value in values:
self._pipe.sadd(table, value)
self._pipe.execute()
else:
return self._redis.sadd(table, values)
def sget(self, table, count = 0, is_pop = True):
datas = []
if is_pop:
count = count if count <= self.sget_count(table) else self.sget_count(table)
if count:
if count > 1:
if not self._is_redis_cluster: self._pipe.multi()
while count:
self._pipe.spop(table)
count -= 1
datas = self._pipe.execute()
else:
datas.append(self._redis.spop(table))
else:
datas = self._redis.srandmember(table, count)
return datas
def sget_count(self, table):
return self._redis.scard(table)
def sdelete(self, table):
'''
@summary: 删除set集合的大键（数据量大的表）
删除大set键，使用sscan命令，每次扫描集合中500个元素，再用srem命令每次删除一个键
若直接用delete命令，会导致Redis阻塞，出现故障切换和应用程序崩溃的故障。
---------
@param table:
---------
@result:
'''
# 当 SCAN 命令的游标参数被设置为 0 时， 服务器将开始一次新的迭代， 而当服务器向用户返回值为 0 的游标时， 表示迭代已结束
cursor = '0'
while cursor != 0:
cursor, data = self._redis.sscan(table, cursor = cursor, count = 10000)
for item in data:
# self._redis.srem(table,item)
self._pipe.srem(table, item)
# print('sdelete %s data size %s'%(table, len(data)))
self._pipe.execute()
def zadd(self, table, values, prioritys = 0):
'''
@summary: 使用有序set集合存储数据， 去重(值存在更新)
---------
@param table:
@param values: 值； 支持list 或 单个值
@param prioritys: 优先级； double类型，支持list 或 单个值。 根据此字段的值来排序, 值越小越优先。 可不传值，默认value的优先级为0
---------
@result:若库中存在 返回0，否则入库，返回1。 批量添加返回None
'''
if isinstance(values, list):
if not isinstance(prioritys, list):
prioritys = [prioritys] * len(values)
else:
assert len(values) == len(prioritys), 'values值要与prioritys值一一对应'
if not self._is_redis_cluster: self._pipe.multi()
for value, priority in zip(values, prioritys):
if self._is_redis_cluster:
self._pipe.zadd(table, priority, value)
else:
self._pipe.zadd(table, value, priority)
self._pipe.execute()
else:
if self._is_redis_cluster:
return self._redis.zadd(table, prioritys, values)
else:
return self._redis.zadd(table, values, prioritys)
def zget(self, table, count = 0, is_pop = True):
'''
@summary: 从有序set集合中获取数据
---------
@param table:
@param count: 数量
@param is_pop:获取数据后，是否在原set集合中删除，默认是
---------
@result: 列表
'''
start_pos = 0 # 包含
end_pos = 0 if count == 0 else count - 1 # 包含
if not self._is_redis_cluster: self._pipe.multi() # 标记事务的开始 参考 http://www.runoob.com/redis/redis-transactions.html
self._pipe.zrange(table, start_pos, end_pos) # 取值
if is_pop: self._pipe.zremrangebyrank(table, start_pos, end_pos) # 删除
results, count = self._pipe.execute()
return results
def zget_count(self, table, priority_min = None, priority_max = None):
'''
@summary: 获取表数据的数量
---------
@param table:
@param priority_min:优先级范围 最小值（包含）
@param priority_max:优先级范围 最大值（包含）
---------
@result:
'''
if priority_min != None and priority_max != None:
return self._redis.zcount(table, priority_min, priority_max)
else:
return self._redis.zcard(table)
def lpush(self, table, values):
if isinstance(values, list):
if not self._is_redis_cluster: self._pipe.multi()
for value in values:
self._pipe.rpush(table, value)
self._pipe.execute()
else:
return self._redis.rpush(table, values)
def lpop(self, table, count = 1):
'''
@summary:
---------
@param table:
@param count:
---------
@result: 返回列表
'''
datas = []
count = count if count <= self.lget_count(table) else self.lget_count(table)
if count:
if count > 1:
if not self._is_redis_cluster: self._pipe.multi()
while count:
data = self._pipe.lpop(table)
count -= 1
datas = self._pipe.execute()
else:
datas.append(self._redis.lpop(table))
return datas
def lget_count(self, table):
return self._redis.llen(table)
def setbit(self, table, offset, value):
self._redis.setbit(table, offset, value)
def getbit(self, table, offset):
return self._redis.getbit(table, offset)
def clear(self, table):
try:
self._redis.delete(table)
except Exception as e:
log.error(e)
class SharQ(object):
"""The SharQ object is the core of this queue.
SharQ does the following.
1. Accepts a configuration file.
2. Initializes the queue.
3. Exposes functions to interact with the queue.
"""
def __init__(self, config_path):
"""Construct a SharQ object by doing the following.
1. Read the configuration path.
2. Load the config.
3. Initialized SharQ.
"""
self.config_path = config_path
self._load_config()
self._initialize()
def _initialize(self):
"""Read the SharQ configuration and set appropriate
variables. Open a redis connection pool and load all
the Lua scripts.
"""
self._key_prefix = self._config.get('redis', 'key_prefix')
self._job_expire_interval = int(
self._config.get('sharq', 'job_expire_interval'))
self._default_job_requeue_limit = int(
self._config.get('sharq', 'default_job_requeue_limit'))
# initalize redis
redis_connection_type = self._config.get('redis', 'conn_type')
db = self._config.get('redis', 'db')
if redis_connection_type == 'unix_sock':
self._r = redis.StrictRedis(db=db,
unix_socket_path=self._config.get(
'redis', 'unix_socket_path'))
elif redis_connection_type == 'tcp_sock':
if self._config.getboolean('redis', 'clustered', fallback=False):
startup_nodes = [{
"host": self._config.get('redis', 'host'),
"port": self._config.get('redis', 'port')
}]
self._r = StrictRedisCluster(startup_nodes=startup_nodes,
decode_responses=True,
skip_full_coverage_check=True)
else:
self._r = redis.StrictRedis(
db=db,
host=self._config.get('redis', 'host'),
port=self._config.get('redis', 'port'))
self._load_lua_scripts()
def _load_config(self):
"""Read the configuration file and load it into memory."""
self._config = ConfigParser.SafeConfigParser()
self._config.read(self.config_path)
def reload_config(self, config_path=None):
"""Reload the configuration from the new config file if provided
else reload the current config file.
"""
if config_path:
self.config_path = config_path
self._load_config()
def _load_lua_scripts(self):
"""Loads all lua scripts required by SharQ."""
# load lua scripts
lua_script_path = os.path.join(
os.path.dirname(os.path.abspath(__file__)), 'scripts/lua')
with open(os.path.join(lua_script_path, 'enqueue.lua'),
'r') as enqueue_file:
self._lua_enqueue_script = enqueue_file.read()
self._lua_enqueue = self._r.register_script(
self._lua_enqueue_script)
with open(os.path.join(lua_script_path, 'dequeue.lua'),
'r') as dequeue_file:
self._lua_dequeue_script = dequeue_file.read()
self._lua_dequeue = self._r.register_script(
self._lua_dequeue_script)
with open(os.path.join(lua_script_path, 'finish.lua'),
'r') as finish_file:
self._lua_finish_script = finish_file.read()
self._lua_finish = self._r.register_script(self._lua_finish_script)
with open(os.path.join(lua_script_path, 'interval.lua'),
'r') as interval_file:
self._lua_interval_script = interval_file.read()
self._lua_interval = self._r.register_script(
self._lua_interval_script)
with open(os.path.join(lua_script_path, 'requeue.lua'),
'r') as requeue_file:
self._lua_requeue_script = requeue_file.read()
self._lua_requeue = self._r.register_script(
self._lua_requeue_script)
with open(os.path.join(lua_script_path, 'metrics.lua'),
'r') as metrics_file:
self._lua_metrics_script = metrics_file.read()
self._lua_metrics = self._r.register_script(
self._lua_metrics_script)
def reload_lua_scripts(self):
"""Lets user reload the lua scripts in run time."""
self._load_lua_scripts()
def enqueue(self,
payload,
interval,
job_id,
queue_id,
queue_type='default',
requeue_limit=None):
"""Enqueues the job into the specified queue_id
of a particular queue_type
"""
# validate all the input
if not is_valid_interval(interval):
raise BadArgumentException('`interval` has an invalid value.')
if not is_valid_identifier(job_id):
raise BadArgumentException('`job_id` has an invalid value.')
if not is_valid_identifier(queue_id):
raise BadArgumentException('`queue_id` has an invalid value.')
if not is_valid_identifier(queue_type):
raise BadArgumentException('`queue_type` has an invalid value.')
if requeue_limit is None:
requeue_limit = self._default_job_requeue_limit
if not is_valid_requeue_limit(requeue_limit):
raise BadArgumentException('`requeue_limit` has an invalid value.')
try:
serialized_payload = serialize_payload(payload)
except TypeError as e:
raise BadArgumentException(e.message)
timestamp = str(generate_epoch())
keys = [self._key_prefix, queue_type]
args = [
timestamp, queue_id, job_id,
'"%s"' % serialized_payload, interval, requeue_limit
]
self._lua_enqueue(keys=keys, args=args)
response = {'status': 'queued'}
return response
def dequeue(self, queue_type='default'):
"""Dequeues a job from any of the ready queues
based on the queue_type. If no job is ready,
returns a failure status.
"""
if not is_valid_identifier(queue_type):
raise BadArgumentException('`queue_type` has an invalid value.')
timestamp = str(generate_epoch())
keys = [self._key_prefix, queue_type]
args = [timestamp, self._job_expire_interval]
dequeue_response = self._lua_dequeue(keys=keys, args=args)
if len(dequeue_response) < 4:
response = {'status': 'failure'}
return response
queue_id, job_id, payload, requeues_remaining = dequeue_response
payload = deserialize_payload(payload[1:-1])
response = {
'status': 'success',
'queue_id': queue_id,
'job_id': job_id,
'payload': payload,
'requeues_remaining': int(requeues_remaining)
}
return response
def finish(self, job_id, queue_id, queue_type='default'):
"""Marks any dequeued job as *completed successfully*.
Any job which gets a finish will be treated as complete
and will be removed from the SharQ.
"""
if not is_valid_identifier(job_id):
raise BadArgumentException('`job_id` has an invalid value.')
if not is_valid_identifier(queue_id):
raise BadArgumentException('`queue_id` has an invalid value.')
if not is_valid_identifier(queue_type):
raise BadArgumentException('`queue_type` has an invalid value.')
keys = [self._key_prefix, queue_type]
args = [queue_id, job_id]
response = {'status': 'success'}
finish_response = self._lua_finish(keys=keys, args=args)
if finish_response == 0:
# the finish failed.
response.update({'status': 'failure'})
return response
def interval(self, interval, queue_id, queue_type='default'):
"""Updates the interval for a specific queue_id
of a particular queue type.
"""
# validate all the input
if not is_valid_interval(interval):
raise BadArgumentException('`interval` has an invalid value.')
if not is_valid_identifier(queue_id):
raise BadArgumentException('`queue_id` has an invalid value.')
if not is_valid_identifier(queue_type):
raise BadArgumentException('`queue_type` has an invalid value.')
# generate the interval key
interval_hmap_key = '%s:interval' % self._key_prefix
interval_queue_key = '%s:%s' % (queue_type, queue_id)
keys = [interval_hmap_key, interval_queue_key]
args = [interval]
interval_response = self._lua_interval(keys=keys, args=args)
if interval_response == 0:
# the queue with the id and type does not exist.
response = {'status': 'failure'}
else:
response = {'status': 'success'}
return response
def requeue(self):
"""Re-queues any expired job (one which does not get an expire
before the job_expiry_interval) back into their respective queue.
This function has to be run at specified intervals to ensure the
expired jobs are re-queued back.
"""
timestamp = str(generate_epoch())
# get all queue_types and requeue one by one.
# not recommended to do this entire process
# in lua as it might take long and block other
# enqueues and dequeues.
active_queue_type_list = self._r.smembers('%s:active:queue_type' %
self._key_prefix)
for queue_type in active_queue_type_list:
# requeue all expired jobs in all queue types.
keys = [self._key_prefix, queue_type]
args = [timestamp]
job_discard_list = self._lua_requeue(keys=keys, args=args)
# discard the jobs if any
for job in job_discard_list:
queue_id, job_id = job.split(':')
# explicitly finishing a job
# is nothing but discard.
self.finish(job_id=job_id,
queue_id=queue_id,
queue_type=queue_type)
def metrics(self, queue_type=None, queue_id=None):
"""Provides a way to get statistics about various parameters like,
* global enqueue / dequeue rates per min.
* per queue enqueue / dequeue rates per min.
* queue length of each queue.
* list of queue ids for each queue type.
"""
if queue_id is not None and not is_valid_identifier(queue_id):
raise BadArgumentException('`queue_id` has an invalid value.')
if queue_type is not None and not is_valid_identifier(queue_type):
raise BadArgumentException('`queue_type` has an invalid value.')
response = {'status': 'failure'}
if not queue_type and not queue_id:
# return global stats.
# list of active queue types (ready + active)
active_queue_types = self._r.smembers('%s:active:queue_type' %
self._key_prefix)
ready_queue_types = self._r.smembers('%s:ready:queue_type' %
self._key_prefix)
all_queue_types = active_queue_types | ready_queue_types
# global rates for past 10 minutes
timestamp = str(generate_epoch())
keys = [self._key_prefix]
args = [timestamp]
enqueue_details, dequeue_details = self._lua_metrics(keys=keys,
args=args)
enqueue_counts = {}
dequeue_counts = {}
# the length of enqueue & dequeue details are always same.
for i in xrange(0, len(enqueue_details), 2):
enqueue_counts[str(enqueue_details[i])] = int(
enqueue_details[i + 1] or 0)
dequeue_counts[str(dequeue_details[i])] = int(
dequeue_details[i + 1] or 0)
response.update({
'status': 'success',
'queue_types': list(all_queue_types),
'enqueue_counts': enqueue_counts,
'dequeue_counts': dequeue_counts
})
return response
elif queue_type and not queue_id:
# return list of queue_ids.
# get data from two sorted sets in a transaction
pipe = self._r.pipeline()
pipe.zrange('%s:%s' % (self._key_prefix, queue_type), 0, -1)
pipe.zrange('%s:%s:active' % (self._key_prefix, queue_type), 0, -1)
ready_queues, active_queues = pipe.execute()
# extract the queue_ids from the queue_id:job_id string
active_queues = [i.split(':')[0] for i in active_queues]
all_queue_set = set(ready_queues) | set(active_queues)
response.update({
'status': 'success',
'queue_ids': list(all_queue_set)
})
return response
elif queue_type and queue_id:
# return specific details.
active_queue_types = self._r.smembers('%s:active:queue_type' %
self._key_prefix)
ready_queue_types = self._r.smembers('%s:ready:queue_type' %
self._key_prefix)
all_queue_types = active_queue_types | ready_queue_types
# queue specific rates for past 10 minutes
timestamp = str(generate_epoch())
keys = ['%s:%s:%s' % (self._key_prefix, queue_type, queue_id)]
args = [timestamp]
enqueue_details, dequeue_details = self._lua_metrics(keys=keys,
args=args)
enqueue_counts = {}
dequeue_counts = {}
# the length of enqueue & dequeue details are always same.
for i in xrange(0, len(enqueue_details), 2):
enqueue_counts[str(enqueue_details[i])] = int(
enqueue_details[i + 1] or 0)
dequeue_counts[str(dequeue_details[i])] = int(
dequeue_details[i + 1] or 0)
# get the queue length for the job queue
queue_length = self._r.llen(
'%s:%s:%s' % (self._key_prefix, queue_type, queue_id))
response.update({
'status': 'success',
'queue_length': int(queue_length),
'enqueue_counts': enqueue_counts,
'dequeue_counts': dequeue_counts
})
return response
elif not queue_type and queue_id:
raise BadArgumentException(
'`queue_id` should be accompanied by `queue_type`.')
return response
def clear_queue(self, queue_type=None, queue_id=None, purge_all=False):
"""clear the all entries in queue with particular queue_id
and queue_type. It takes an optional argument,
purge_all : if True, then it will remove the related resources
from the redis.
"""
if queue_id is None or not is_valid_identifier(queue_id):
raise BadArgumentException('`queue_id` has an invalid value.')
if queue_type is None or not is_valid_identifier(queue_type):
raise BadArgumentException('`queue_type` has an invalid value.')
response = {'status': 'Failure', 'message': 'No queued calls found'}
# remove from the primary sorted set
primary_set = '{}:{}'.format(self._key_prefix, queue_type)
queued_status = self._r.zrem(primary_set, queue_id)
if queued_status:
response.update({
'status': 'Success',
'message': 'Successfully removed all queued calls'
})
# do a full cleanup of reources
# although this is not necessary as we don't remove resources
# while dequeue operation
job_queue_list = '{}:{}:{}'.format(self._key_prefix, queue_type,
queue_id)
if queued_status and purge_all:
job_list = self._r.lrange(job_queue_list, 0, -1)
pipe = self._r.pipeline()
# clear the payload data for job_uuid
for job_uuid in job_list:
if job_uuid is None:
continue
payload_set = '{}:payload'.format(self._key_prefix)
job_payload_key = '{}:{}:{}'.format(queue_type, queue_id,
job_uuid)
pipe.hdel(payload_set, job_payload_key)
# clear jobrequest interval
interval_set = '{}:interval'.format(self._key_prefix)
job_interval_key = '{}:{}'.format(queue_type, queue_id)
pipe.hdel(interval_set, job_interval_key)
# clear job_queue_list
pipe.delete(job_queue_list)
pipe.execute()
response.update({
'status':
'Success',
'message':
'Successfully removed all queued calls and purged related resources'
})
else:
# always delete the job queue list
self._r.delete(job_queue_list)
return response
class RedisDB:
def __init__(self,
ip_ports=None,
db=None,
user_pass=None,
url=None,
decode_responses=True,
service_name=None,
max_connections=32,
**kwargs):
"""
redis的封装
Args:
ip_ports: ip:port 多个可写为列表或者逗号隔开 如 ip1:port1,ip2:port2 或 ["ip1:port1", "ip2:port2"]
db:
user_pass:
url:
decode_responses:
service_name: 适用于redis哨兵模式
"""
# 可能会改setting中的值，所以此处不能直接赋值为默认值，需要后加载赋值
if ip_ports is None:
ip_ports = setting.REDISDB_IP_PORTS
if db is None:
db = setting.REDISDB_DB
if user_pass is None:
user_pass = setting.REDISDB_USER_PASS
if service_name is None:
service_name = setting.REDISDB_SERVICE_NAME
self._is_redis_cluster = False
try:
if not url:
ip_ports = (ip_ports if isinstance(ip_ports, list) else
ip_ports.split(","))
if len(ip_ports) > 1:
startup_nodes = []
for ip_port in ip_ports:
ip, port = ip_port.split(":")
startup_nodes.append({"host": ip, "port": port})
if service_name:
log.debug("使用redis哨兵模式")
hosts = [(node["host"], node["port"])
for node in startup_nodes]
sentinel = Sentinel(hosts, socket_timeout=3, **kwargs)
self._redis = sentinel.master_for(
service_name,
password=user_pass,
db=db,
redis_class=redis.StrictRedis,
decode_responses=decode_responses,
max_connections=max_connections,
**kwargs)
else:
log.debug("使用redis集群模式")
self._redis = StrictRedisCluster(
startup_nodes=startup_nodes,
decode_responses=decode_responses,
password=user_pass,
max_connections=max_connections,
**kwargs)
self._is_redis_cluster = True
else:
ip, port = ip_ports[0].split(":")
self._redis = redis.StrictRedis(
host=ip,
port=port,
db=db,
password=user_pass,
decode_responses=decode_responses,
max_connections=max_connections,
**kwargs)
else:
self._redis = redis.StrictRedis.from_url(
url, decode_responses=decode_responses)
except Exception as e:
raise
else:
if not url:
log.debug("连接到redis数据库 %s db%s" % (ip_ports, db))
else:
log.debug("连接到redis数据库 %s" % (url))
self._ip_ports = ip_ports
self._db = db
self._user_pass = user_pass
self._url = url
def __repr__(self):
if self._url:
return "<Redisdb url:{}>".format(self._url)
return "<Redisdb ip_ports: {} db:{} user_pass:{}>".format(
self._ip_ports, self._db, self._user_pass)
@classmethod
def from_url(cls, url):
return cls(url=url)
def sadd(self, table, values):
"""
@summary: 使用无序set集合存储数据， 去重
---------
@param table:
@param values: 值； 支持list 或 单个值
---------
@result: 若库中存在 返回0，否则入库，返回1。 批量添加返回None
"""
if isinstance(values, list):
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi()
for value in values:
pipe.sadd(table, value)
pipe.execute()
else:
return self._redis.sadd(table, values)
def sget(self, table, count=1, is_pop=True):
"""
返回 list 如 ['1'] 或 []
@param table:
@param count:
@param is_pop:
@return:
"""
datas = []
if is_pop:
count = count if count <= self.sget_count(
table) else self.sget_count(table)
if count:
if count > 1:
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi()
while count:
pipe.spop(table)
count -= 1
datas = pipe.execute()
else:
datas.append(self._redis.spop(table))
else:
datas = self._redis.srandmember(table, count)
return datas
def srem(self, table, values):
"""
@summary: 移除集合中的指定元素
---------
@param table:
@param values: 一个或者列表
---------
@result:
"""
if isinstance(values, list):
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi()
for value in values:
pipe.srem(table, value)
pipe.execute()
else:
self._redis.srem(table, values)
def sget_count(self, table):
return self._redis.scard(table)
def sdelete(self, table):
"""
@summary: 删除set集合的大键（数据量大的表）
删除大set键，使用sscan命令，每次扫描集合中500个元素，再用srem命令每次删除一个键
若直接用delete命令，会导致Redis阻塞，出现故障切换和应用程序崩溃的故障。
---------
@param table:
---------
@result:
"""
# 当 SCAN 命令的游标参数被设置为 0 时， 服务器将开始一次新的迭代， 而当服务器向用户返回值为 0 的游标时， 表示迭代已结束
cursor = "0"
while cursor != 0:
cursor, data = self._redis.sscan(table, cursor=cursor, count=500)
for item in data:
# pipe.srem(table, item)
self._redis.srem(table, item)
# pipe.execute()
def sismember(self, table, key):
"Return a boolean indicating if ``value`` is a member of set ``name``"
return self._redis.sismember(table, key)
def zadd(self, table, values, prioritys=0):
"""
@summary: 使用有序set集合存储数据， 去重(值存在更新)
---------
@param table:
@param values: 值； 支持list 或 单个值
@param prioritys: 优先级； double类型，支持list 或 单个值。 根据此字段的值来排序, 值越小越优先。 可不传值，默认value的优先级为0
---------
@result:若库中存在 返回0，否则入库，返回1。 批量添加返回 [0, 1 ...]
"""
if isinstance(values, list):
if not isinstance(prioritys, list):
prioritys = [prioritys] * len(values)
else:
assert len(values) == len(prioritys), "values值要与prioritys值一一对应"
pipe = self._redis.pipeline(transaction=True)
if not self._is_redis_cluster:
pipe.multi()
for value, priority in zip(values, prioritys):
pipe.zadd(table, priority, value)
return pipe.execute()
else:
return self._redis.zadd(table, prioritys, values)
def zget(self, table, count=1, is_pop=True):
"""
@summary: 从有序set集合中获取数据 优先返回分数小的（优先级高的）
---------
@param table:
@param count: 数量 -1 返回全部数据
@param is_pop:获取数据后，是否在原set集合中删除，默认是
---------
@result: 列表
"""
start_pos = 0 # 包含
end_pos = count - 1 if count > 0 else count
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi(
) # 标记事务的开始 参考 http://www.runoob.com/redis/redis-transactions.html
pipe.zrange(table, start_pos, end_pos) # 取值
if is_pop:
pipe.zremrangebyrank(table, start_pos, end_pos) # 删除
results, *count = pipe.execute()
return results
def zremrangebyscore(self, table, priority_min, priority_max):
"""
根据分数移除成员 闭区间
@param table:
@param priority_min:
@param priority_max:
@return: 被移除的成员个数
"""
return self._redis.zremrangebyscore(table, priority_min, priority_max)
def zrangebyscore(self,
table,
priority_min,
priority_max,
count=None,
is_pop=True):
"""
@summary: 返回指定分数区间的数据 闭区间
---------
@param table:
@param priority_min: 优先级越小越优先
@param priority_max:
@param count: 获取的数量，为空则表示分数区间内的全部数据
@param is_pop: 是否删除
---------
@result:
"""
# 使用lua脚本， 保证操作的原子性
lua = """
local key = KEYS[1]
local min_score = ARGV[2]
local max_score = ARGV[3]
local is_pop = ARGV[4]
local count = ARGV[5]
-- 取值
local datas = nil
if count then
datas = redis.call('zrangebyscore', key, min_score, max_score, 'limit', 0, count)
else
datas = redis.call('zrangebyscore', key, min_score, max_score)
end
-- 删除redis中刚取到的值
if (is_pop) then
for i=1, #datas do
redis.call('zrem', key, datas[i])
end
end
return datas
"""
cmd = self._redis.register_script(lua)
if count:
res = cmd(keys=[table],
args=[table, priority_min, priority_max, is_pop, count])
else:
res = cmd(keys=[table],
args=[table, priority_min, priority_max, is_pop])
return res
def zrangebyscore_increase_score(self,
table,
priority_min,
priority_max,
increase_score,
count=None):
"""
@summary: 返回指定分数区间的数据 闭区间， 同时修改分数
---------
@param table:
@param priority_min: 最小分数
@param priority_max: 最大分数
@param increase_score: 分数值增量 正数则在原有的分数上叠加，负数则相减
@param count: 获取的数量，为空则表示分数区间内的全部数据
---------
@result:
"""
# 使用lua脚本， 保证操作的原子性
lua = """
local key = KEYS[1]
local min_score = ARGV[1]
local max_score = ARGV[2]
local increase_score = ARGV[3]
local count = ARGV[4]
-- 取值
local datas = nil
if count then
datas = redis.call('zrangebyscore', key, min_score, max_score, 'limit', 0, count)
else
datas = redis.call('zrangebyscore', key, min_score, max_score)
end
--修改优先级
for i=1, #datas do
redis.call('zincrby', key, increase_score, datas[i])
end
return datas
"""
cmd = self._redis.register_script(lua)
if count:
res = cmd(keys=[table],
args=[priority_min, priority_max, increase_score, count])
else:
res = cmd(keys=[table],
args=[priority_min, priority_max, increase_score])
return res
def zrangebyscore_set_score(self,
table,
priority_min,
priority_max,
score,
count=None):
"""
@summary: 返回指定分数区间的数据 闭区间， 同时修改分数
---------
@param table:
@param priority_min: 最小分数
@param priority_max: 最大分数
@param score: 分数值
@param count: 获取的数量，为空则表示分数区间内的全部数据
---------
@result:
"""
# 使用lua脚本， 保证操作的原子性
lua = """
local key = KEYS[1]
local min_score = ARGV[1]
local max_score = ARGV[2]
local set_score = ARGV[3]
local count = ARGV[4]
-- 取值
local datas = nil
if count then
datas = redis.call('zrangebyscore', key, min_score, max_score, 'withscores','limit', 0, count)
else
datas = redis.call('zrangebyscore', key, min_score, max_score, 'withscores')
end
local real_datas = {} -- 数据
--修改优先级
for i=1, #datas, 2 do
local data = datas[i]
local score = datas[i+1]
table.insert(real_datas, data) -- 添加数据
redis.call('zincrby', key, set_score - score, datas[i])
end
return real_datas
"""
cmd = self._redis.register_script(lua)
if count:
res = cmd(keys=[table],
args=[priority_min, priority_max, score, count])
else:
res = cmd(keys=[table], args=[priority_min, priority_max, score])
return res
def zget_count(self, table, priority_min=None, priority_max=None):
"""
@summary: 获取表数据的数量
---------
@param table:
@param priority_min:优先级范围 最小值（包含）
@param priority_max:优先级范围 最大值（包含）
---------
@result:
"""
if priority_min != None and priority_max != None:
return self._redis.zcount(table, priority_min, priority_max)
else:
return self._redis.zcard(table)
def zrem(self, table, values):
"""
@summary: 移除集合中的指定元素
---------
@param table:
@param values: 一个或者列表
---------
@result:
"""
if isinstance(values, list):
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi()
for value in values:
pipe.zrem(table, value)
pipe.execute()
else:
self._redis.zrem(table, values)
def zexists(self, table, values):
"""
利用zscore判断某元素是否存在
@param values:
@return:
"""
is_exists = []
if isinstance(values, list):
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
pipe.multi()
for value in values:
pipe.zscore(table, value)
is_exists_temp = pipe.execute()
for is_exist in is_exists_temp:
if is_exist != None:
is_exists.append(1)
else:
is_exists.append(0)
else:
is_exists = self._redis.zscore(table, values)
is_exists = 1 if is_exists != None else 0
return is_exists
def lpush(self, table, values):
if isinstance(values, list):
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi()
for value in values:
pipe.rpush(table, value)
pipe.execute()
else:
return self._redis.rpush(table, values)
def lpop(self, table, count=1):
"""
@summary:
---------
@param table:
@param count:
---------
@result: count>1时返回列表
"""
datas = None
count = count if count <= self.lget_count(table) else self.lget_count(
table)
if count:
if count > 1:
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
if not self._is_redis_cluster:
pipe.multi()
while count:
pipe.lpop(table)
count -= 1
datas = pipe.execute()
else:
datas = self._redis.lpop(table)
return datas
def rpoplpush(self, from_table, to_table=None):
"""
将列表 from_table 中的最后一个元素(尾元素)弹出，并返回给客户端。
将 from_table 弹出的元素插入到列表 to_table ，作为 to_table 列表的的头元素。
如果 from_table 和 to_table 相同，则列表中的表尾元素被移动到表头，并返回该元素，可以把这种特殊情况视作列表的旋转(rotation)操作
@param from_table:
@param to_table:
@return:
"""
if not to_table:
to_table = from_table
return self._redis.rpoplpush(from_table, to_table)
def lget_count(self, table):
return self._redis.llen(table)
def lrem(self, table, value, num=0):
return self._redis.lrem(table, value, num)
def hset(self, table, key, value):
"""
@summary:
如果 key 不存在，一个新的哈希表被创建并进行 HSET 操作。
如果域 field 已经存在于哈希表中，旧值将被覆盖
---------
@param table:
@param key:
@param value:
---------
@result: 1 新插入； 0 覆盖
"""
return self._redis.hset(table, key, value)
def hset_batch(self, table, datas):
"""
批量插入
Args:
datas:
[[key, value]]
Returns:
"""
pipe = self._redis.pipeline(transaction=True)
if not self._is_redis_cluster:
pipe.multi()
for key, value in datas:
pipe.hset(table, key, value)
return pipe.execute()
def hincrby(self, table, key, increment):
return self._redis.hincrby(table, key, increment)
def hget(self, table, key, is_pop=False):
if not is_pop:
return self._redis.hget(table, key)
else:
lua = """
local key = KEYS[1]
local field = ARGV[1]
-- 取值
local datas = redis.call('hget', key, field)
-- 删除值
redis.call('hdel', key, field)
return datas
"""
cmd = self._redis.register_script(lua)
res = cmd(keys=[table], args=[key])
return res
def hgetall(self, table):
return self._redis.hgetall(table)
def hexists(self, table, key):
return self._redis.hexists(table, key)
def hdel(self, table, *keys):
"""
@summary: 删除对应的key 可传多个
---------
@param table:
@param *keys:
---------
@result:
"""
self._redis.hdel(table, *keys)
def hget_count(self, table):
return self._redis.hlen(table)
def setbit(self, table, offsets, values):
"""
设置字符串数组某一位的值， 返回之前的值
@param table:
@param offsets: 支持列表或单个值
@param values: 支持列表或单个值
@return: list / 单个值
"""
if isinstance(offsets, list):
if not isinstance(values, list):
values = [values] * len(offsets)
else:
assert len(offsets) == len(values), "offsets值要与values值一一对应"
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
pipe.multi()
for offset, value in zip(offsets, values):
pipe.setbit(table, offset, value)
return pipe.execute()
else:
return self._redis.setbit(table, offsets, values)
def getbit(self, table, offsets):
"""
取字符串数组某一位的值
@param table:
@param offsets: 支持列表
@return: list / 单个值
"""
if isinstance(offsets, list):
pipe = self._redis.pipeline(
transaction=True
) # redis-py默认在执行每次请求都会创建（连接池申请连接）和断开（归还连接池）一次连接操作，如果想要在一次请求中指定多个命令，则可以使用pipline实现一次请求指定多个命令，并且默认情况下一次pipline 是原子性操作。
pipe.multi()
for offset in offsets:
pipe.getbit(table, offset)
return pipe.execute()
else:
return self._redis.getbit(table, offsets)
def bitcount(self, table):
return self._redis.bitcount(table)
def strset(self, table, value, **kwargs):
return self._redis.set(table, value, **kwargs)
def str_incrby(self, table, value):
return self._redis.incrby(table, value)
def strget(self, table):
return self._redis.get(table)
def strlen(self, table):
return self._redis.strlen(table)
def getkeys(self, regex):
return self._redis.keys(regex)
def exists_key(self, key):
return self._redis.exists(key)
def set_expire(self, key, seconds):
"""
@summary: 设置过期时间
---------
@param key:
@param seconds: 秒
---------
@result:
"""
self._redis.expire(key, seconds)
def clear(self, table):
try:
self._redis.delete(table)
except Exception as e:
log.error(e)
def get_redis_obj(self):
return self._redis
class RedisClient(object):
def __init__(self, key, startup_nodes):
"""
init cluster
"""
self.key = key
self.conn = StrictRedisCluster(startup_nodes=startup_nodes,
decode_responses=True)
def hdel(self, field):
"""
delete an item
:param field:
:return:
"""
self.conn.hdel(self.key, field)
def hexists(self, field):
"""
判断 key 中是否含有 field
:param field:
:return:
"""
return self.conn.hexists(self.key, field)
def hget(self, field):
"""
返回key中指定 field 中的 value
:param field:
:return:
"""
value = self.conn.hget(self.key, field)
if isinstance(value, bytes):
return value.decode('utf-8')
else:
return value if value else None
def hgetall(self):
"""
获取 {filed: value, field1: value1....}
:return:
"""
all_dict = self.conn.hgetall(self.key)
if not all_dict:
return
elif sys.version_info.major == 3:
return {
field.decode('utf-8'): value.decode('utf-8')
for field, value in all_dict.items()
}
else:
return all_dict
def hkeys(self):
"""
获取key中所有field
:return:
"""
field = self.conn.hkeys(self.key)
if isinstance(field, bytes):
return field.decode('utf-8')
else:
return field if field else None
def hlen(self):
"""
获取所有 filed 数量
:return:
"""
return self.conn.hlen(self.key)
def hset(self, field, value):
"""
设置 field: value
:param field:
:param value:
:return:
"""
self.conn.hset(self.key, field, value)
def hvals(self):
"""
获取所有values
:return:
"""
values = self.conn.hvals(self.key)
if not values:
return
elif sys.version_info.major == 3:
return [value.decode('utf-8') for value in values]
else:
return values
def change_key(self, key):
"""
替换 key
:param key:
:return:
"""
self.key = key
# ===============================================
def blpop(self, timeout):
self.conn.blpop(self.key, timeout=timeout)
def brpop(self, timeout):
self.conn.brpop(self.key, timeout=timeout)
def brpoplpush(self, dst, timeout):
self.conn.brpoplpush(self.key, dst=dst, timeout=timeout)
def lindex(self, i):
self.conn.lindex(self.key, index=i)
def llen(self):
self.conn.llen(self.key)
def lpop(self):
self.conn.lpop(self.key)
def lpush(self):
self.conn.lpush(self.key)
def lrange(self, start, stop):
self.conn.lrange(self.key, start, stop)
def lset(self, i, value):
self.conn.lset(self.key, index=i, value=value)
def rpop(self):
self.conn.rpop(self.key)
def rpoplpush(self, dst):
self.conn.rpoplpush(self.key, dst=dst)
def rpush(self, value):
self.conn.rpush(self.key, value)
class RedisMiddleware(object):
"""
任务管理器，负责任务相关操作，如校验是否新增，读取已抓取任务文本
"""
def __init__(self, redis_params):
self.redis_cli = StrictRedisCluster(
startup_nodes=redis_params.get('startup_nodes', ''),
password=redis_params.get('password', ''))
self.bloom_filter = BloomFilter(
self.redis_cli, blockNum=5,
key='bloomfilter_weibo') # url的过滤器，分6个块存,内存空间默认512M
def redis_del(self, key=None):
"""
删除redis对应的键
目前用在循环抓取时候，清空列表url，
列表url每次循环只抓取一遍，直至下次循环
:return:
"""
if not key:
return
res = self.redis_cli.delete(key)
return res
def redis_rpush(self, name, data):
"""
推入数据到redis指定任务列表中
rpush,将新的数据放在最后面
:return:
"""
try:
if isinstance(data, list):
for each in data:
self.redis_cli.rpush(name, each)
else:
self.redis_cli.lpush(name, data)
except:
return
def redis_lpush(self, name, data):
"""
推入数据到redis指定任务列表中
lpush,将新的数据放在最前面
:return:
"""
try:
if isinstance(data, list):
for each in data:
self.redis_cli.lpush(name, each)
else:
self.redis_cli.lpush(name, data)
except:
return
def redis_rpop(self, name):
"""
从指定任务列表中获取数据
rpop,从最后取
:return:
"""
try:
res = self.redis_cli.rpop(name)
return res
except:
return
def redis_lpop(self, name):
"""
从指定任务列表中获取数据
lpop,从头部取
:return:
"""
try:
res = self.redis_cli.lpop(name)
return res
except:
return
def redis_brpop(self, name, timeout=1):
"""
从指定任务列表中获取数据
brpop,阻塞，从最后取
:return:
"""
try:
unuse, res = self.redis_cli.brpop(name, timeout=timeout)
return res
except Exception as e:
print(e)
return
def redis_query(self, name):
"""
查询指定任务列表中数据
:param name:
:return:
"""
try:
res = self.redis_cli.llen(name)
return res
except:
return
def redis_sadd(self, name, data):
"""
集合中插入数据
:return:
"""
try:
if isinstance(data, list) or isinstance(data, set):
for each in data:
self.redis_cli.sadd(name, each)
else:
self.redis_cli.sadd(name, data)
except:
return
def redis_sismember(self, name, data):
"""
校验元素是否存在于集合中
:return:
"""
return self.redis_cli.sismember(name, data)
def redis_scard(self, name):
"""
返回集合成员个数
:return:
"""
return int(self.redis_cli.scard(name))
def redis_spop(self, name):
"""
获取集合中的随机一个元素
:param name:
:return:
"""
return self.redis_cli.spop(name)
def redis_srem(self, name, data):
"""
移除指定成员
:param name:
:param data:
:return:
"""
self.redis_cli.srem(name, data)
THREAD_PROXY_MAP = {} # 线程与代理关系
season = time.strftime('%Y-%m') # 用户的玩家列表url中season参数
def get_redis_proxy():
'''
从redis相应的key中获取代理ip(读取快代理的代理ip)
:return:
'''
current_time = int(time.strftime('%H%M%S'))
if 001000 >= current_time >= 000000: # 退出进程
print time.strftime('[%Y-%m-%d %H:%M:%S]：'), 'get_redis_proxy()退出'
return False
startup_nodes = [{'host': 'redis3', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
pubg_friends_proxy_length = r.llen(
'spider:pubg_friends:proxy:kuai') # pubg_friends
print time.strftime(
'[%Y-%m-%d %H:%M:%S]'
), 'redis中pubg_friends的代理ip长度：', pubg_friends_proxy_length
if pubg_friends_proxy_length >= 50:
proxy_length = 50
else:
print '当前快代理redis中的代理数量少于50：', r.llen('spider:pubg_friends:proxy:kuai')
time.sleep(60)
return get_redis_proxy()
for i in xrange(proxy_length):
ip = r.lpop('spider:pubg_friends:proxy:kuai')
if ip:
proxies = {
'http': "http://{ip}".format(ip=ip),
PROXY_IP_Q = Queue.Queue() # 代理ip队列
season = time.strftime('%Y-%m') # 用户的玩家列表url中season参数
def get_redis_proxy():
'''
从redis相应的key中获取代理ip(读取快代理的代理ip)
:return:
'''
current_time = int(time.strftime('%H%M%S'))
if 001000 >= current_time >= 000000: # 退出进程
print time.strftime('[%Y-%m-%d %H:%M:%S]：'), 'get_redis_proxy()退出'
return False
startup_nodes = [{'host': 'redis3', 'port': '6379'}]
r = StrictRedisCluster(startup_nodes=startup_nodes, decode_responses=True)
pubg_death_proxy_length = r.llen(
'spider:pubg_death:proxy:kuai') # pubg_death
print time.strftime('[%Y-%m-%d %H:%M:%S]'
), 'redis中pubg_death的代理ip长度：', pubg_death_proxy_length
if pubg_death_proxy_length >= 50:
proxy_length = 50
else:
print '当前快代理redis中的代理数量少于50：', r.llen('spider:pubg_death:proxy:kuai')
time.sleep(60)
return get_redis_proxy()
for i in xrange(proxy_length):
ip = r.lpop('spider:pubg_death:proxy:kuai')
if ip:
proxies = {
'http': "http://{ip}".format(ip=ip),
'https': "http://{ip}".format(ip=ip)
#!/usr/bin/python3.4
# -*- coding: utf-8 -*-
import redis
from rediscluster import StrictRedisCluster
redis_nodes = [{'host': '192.168.230.218', 'port': 6380},
{'host': '192.168.230.218', 'port': 6381},
{'host': '192.168.230.218', 'port': 6382},
{'host': '192.168.230.223', 'port': 6383},
{'host': '192.168.230.223', 'port': 6384},
{'host': '192.168.230.223', 'port': 6385}
]
r = StrictRedisCluster(startup_nodes=redis_nodes)
name = "url"
length = r.llen(name)
print(length)
print(r.lrange(name, 0, -1))
class RedisMiddleware(object):
"""
任务管理器，负责任务相关操作，如校验是否新增，读取已抓取任务文本
"""
def __init__(self, taskname, redis_params):
# self._mkdata()
self.redis_cli = StrictRedisCluster(
startup_nodes=redis_params.get('startup_nodes', ''),
password=redis_params.get('password', ''))
# 实例化两个bloomfilter
self.bloom_urls = BloomFilter(
self.redis_cli, blockNum=6,
key='bloomfilter_pub') # url的过滤器，分6个块存,内存空间默认512M
# list的过滤器，默认1个块存,内存空间给32M
self.bloom_list = BloomFilter(self.redis_cli,
key='{}:redis_list'.format(taskname),
bit_size=1 << 28)
# self.redis_cli = redis.Redis(host=redis_host, port=redis_port, db=0, password=redis_psw)
def redis_del(self, key=None):
"""
删除redis对应的键
目前用在循环抓取时候，清空列表url，
列表url每次循环只抓取一遍，直至下次循环
:return:
"""
if not key:
return
res = self.redis_cli.delete(key)
return res
def redis_push(self, name, data):
"""
推入数据到redis指定任务列表中
lpush,将新的数据放在最前面
:return:
"""
try:
if isinstance(data, list):
for each in data:
self.redis_cli.lpush(name, each)
else:
self.redis_cli.lpush(name, data)
except:
return
def redis_pop(self, name):
"""
从指定任务列表中获取数据
rpop,从最后取
:return:
"""
try:
res = self.redis_cli.rpop(name)
return res
except:
return
def redis_brpop(self, name, timeout=1):
"""
从指定任务列表中获取数据
brpop,阻塞，从最后取
:return:
"""
try:
unuse, res = self.redis_cli.brpop(name, timeout=timeout)
return res
except Exception as e:
print(e)
return
def redis_query(self, name):
"""
查询指定任务列表中数据
:param name:
:return:
"""
try:
res = self.redis_cli.llen(name)
return res
except:
return
class RedisQueue(object):
"""
A Queue like message built over redis
"""
Empty = BaseQueue.Empty
Full = BaseQueue.Full
max_timeout = 0.3
def __init__(self, name, host='localhost', port=6379, db=0,
maxsize=0, lazy_limit=True, password=None, cluster_nodes=None):
"""
Constructor for RedisQueue
maxsize: an integer that sets the upperbound limit on the number of
items that can be placed in the queue.
lazy_limit: redis queue is shared via instance, a lazy size limit is used
for better performance.
"""
self.name = name
if(cluster_nodes is not None):
from rediscluster import StrictRedisCluster
self.redis = StrictRedisCluster(startup_nodes=cluster_nodes)
else:
self.redis = redis.StrictRedis(host=host, port=port, db=db, password=password)
self.maxsize = maxsize
self.lazy_limit = lazy_limit
self.last_qsize = 0
def qsize(self):
self.last_qsize = self.redis.llen(self.name)
return self.last_qsize
def empty(self):
if self.qsize() == 0:
return True
else:
return False
def full(self):
if self.maxsize and self.qsize() >= self.maxsize:
return True
else:
return False
def put_nowait(self, obj):
if self.lazy_limit and self.last_qsize < self.maxsize:
pass
elif self.full():
raise self.Full
self.last_qsize = self.redis.rpush(self.name, umsgpack.packb(obj))
return True
def put(self, obj, block=True, timeout=None):
if not block:
return self.put_nowait(obj)
start_time = time.time()
while True:
try:
return self.put_nowait(obj)
except self.Full:
if timeout:
lasted = time.time() - start_time
if timeout > lasted:
time.sleep(min(self.max_timeout, timeout - lasted))
else:
raise
else:
time.sleep(self.max_timeout)
def get_nowait(self):
ret = self.redis.lpop(self.name)
if ret is None:
raise self.Empty
return umsgpack.unpackb(ret)
def get(self, block=True, timeout=None):
if not block:
return self.get_nowait()
start_time = time.time()
while True:
try:
return self.get_nowait()
except self.Empty:
if timeout:
lasted = time.time() - start_time
if timeout > lasted:
time.sleep(min(self.max_timeout, timeout - lasted))
else:
raise
else:
time.sleep(self.max_timeout)
class RedisQueue(object):
"""
A Queue like message built over redis
"""
Empty = BaseQueue.Empty
Full = BaseQueue.Full
max_timeout = 0.3
def __init__(self, name, host='localhost', port=6379, db=0,
maxsize=0, lazy_limit=True, password=None, cluster_nodes=None):
"""
Constructor for RedisQueue
maxsize: an integer that sets the upperbound limit on the number of
items that can be placed in the queue.
lazy_limit: redis queue is shared via instance, a lazy size limit is used
for better performance.
"""
self.name = name
if(cluster_nodes is not None):
from rediscluster import StrictRedisCluster
self.redis = StrictRedisCluster(startup_nodes=cluster_nodes)
else:
self.redis = redis.StrictRedis(host=host, port=port, db=db, password=password)
self.maxsize = maxsize
self.lazy_limit = lazy_limit
self.last_qsize = 0
def qsize(self):
self.last_qsize = self.redis.llen(self.name)
return self.last_qsize
def empty(self):
if self.qsize() == 0:
return True
else:
return False
def full(self):
if self.maxsize and self.qsize() >= self.maxsize:
return True
else:
return False
def put_nowait(self, obj):
if self.lazy_limit and self.last_qsize < self.maxsize:
pass
elif self.full():
raise self.Full
self.last_qsize = self.redis.rpush(self.name, umsgpack.packb(obj))
return True
def put(self, obj, block=True, timeout=None):
if not block:
return self.put_nowait(obj)
start_time = time.time()
while True:
try:
return self.put_nowait(obj)
except self.Full:
if timeout:
lasted = time.time() - start_time
if timeout > lasted:
time.sleep(min(self.max_timeout, timeout - lasted))
else:
raise
else:
time.sleep(self.max_timeout)
def get_nowait(self):
ret = self.redis.lpop(self.name)
if ret is None:
raise self.Empty
return umsgpack.unpackb(ret)
def get(self, block=True, timeout=None):
if not block:
return self.get_nowait()
start_time = time.time()
while True:
try:
return self.get_nowait()
except self.Empty:
if timeout:
lasted = time.time() - start_time
if timeout > lasted:
time.sleep(min(self.max_timeout, timeout - lasted))
else:
raise
else:
time.sleep(self.max_timeout)
