class SSDBHashMap(object, SSDBBase):
def __init__(self,
hashmap_name,
host="127.0.0.1",
port=8888,
max_connections=10,
timeout=60):
self.hashmap_name = hashmap_name
self.host = host
self.port = port
self.max_connections = max_connections
self.timeout = timeout
pool = BlockingConnectionPool(connection_class=Connection,
max_connections=max_connections,
timeout=timeout,
host=host,
port=port)
self.ssdb = SSDB(connection_pool=pool)
SSDBBase.__init__(self, self.ssdb)
def set(self, key, value):
return self.ssdb.hset(self.hashmap_name, key, value)
def get(self, key):
return self.ssdb.hget(self.hashmap_name, key)
def delete(self, key):
return self.ssdb.hdel(self.hashmap_name, key)
def keys(self, name_start="", name_end="", limit=10):
return self.ssdb.hkeys(self.hashmap_name, name_start, name_end, limit)
def exists(self, key):
return self.ssdb.hexists(self.hashmap_name, key)
def size(self):
return self.ssdb.hsize(self.hashmap_name)
def list(self, name_start="", name_end="", limit=10):
#列出名字处于区间 (name_start, name_end] 的 hashmap
return self.ssdb.hlist(name_start, name_end, limit)
def scan(self, key_start, key_end="", limit=10):
return self.ssdb.hscan(self.hashmap_name, key_start, key_end, limit)
def clear(self):
return self.ssdb.hclear(self.hashmap_name)
class ssdb_proxy(object):
def __init__(self, set_name, host, port, max_connections=2, timeout=60):
"""数据库初始化"""
self.set_name = set_name
self.host = host
self.port = port
self.max_connections = max_connections
self.timeout = timeout
pool = BlockingConnectionPool(connection_class=Connection,
max_connections=max_connections,
timeout=timeout,
host=host,
port=port)
self.ssdb = SSDB(connection_pool=pool)
pass
def hset(self, key, value):
"""添加集合"""
return self.ssdb.hset(self.set_name, key, value)
def hget(self, key):
"""获取key所对应的值"""
return self.ssdb.hget(self.set_name, key)
def hgetall(self):
"""获取所有的数据"""
return self.ssdb.hgetall(self.set_name)
def hdel(self, key):
"""删除数据"""
return self.ssdb.hdel(self.set_name, key)
def hexists(self, key):
"""判断Key是否存在"""
return self.ssdb.hexists(self.set_name, key)
def size(self):
"""获取结合的大小"""
return self.ssdb.hsize(self.set_name)
def clean(self):
"""清楚所有数据"""
return self.ssdb.hclear(self.set_name)
# Get URLS from SSDB and turn into a text file called links
import subprocess
import os
from ssdb import SSDB
ssdb = SSDB(host='104.131.54.255', port=8888)
# get hostid
hostid = subprocess.check_output(['curl','http://ipinfo.io/ip']).strip()
limit = ssdb.hget("crawl","limit")
offset = ssdb.hget("crawl", hostid)
users = ssdb.zrange("instagram-users", int(offset), int(limit))
arr = users
f = open("links", "w")
f.write("\n".join(map(lambda x: str(x), arr)))
f.close()
class SsdbClient(object):
"""
SSDB client
SSDB中代理存放的容器为hash：
原始代理存放在name为raw_proxy的hash中，key为代理的ip:port，value为None,以后扩展可能会加入代理属性；
验证后供flask使用的代理存放在name为useful_proxy的hash中，key为代理的ip:port，value为None,以后扩展可能会加入代理属性；
"""
def __init__(self, name, host, port):
"""
init
:param name: hash name
:param host: ssdb host
:param port: ssdb port
:return:
"""
self.name = name
self.__conn = SSDB(
connection_pool=BlockingConnectionPool(host=host, port=port))
def get(self):
"""
get an item
从useful_proxy_queue随机获取一个可用代理, 使用前需要调用changeTable("useful_proxy_queue")
:return:
"""
values = self.__conn.hgetall(name=self.name)
return random.choice(values.keys()) if values else None
def put(self, key):
"""
put an item
将代理放入hash, 使用changeTable指定hash name
:param key:
:return:
"""
key = json.dump(key, ensure_ascii=False).encode('utf-8') if isinstance(
key, (dict, list)) else key
return self.__conn.hincr(self.name, key, 1)
# return self.__conn.hset(self.name, value, None)
def getvalue(self, key):
value = self.__conn.hget(self.name, key)
return value if value else None
def pop(self):
"""
pop an item
弹出一个代理， 使用changeTable指定hash name
:return:
"""
key = self.get()
if key:
self.__conn.hdel(self.name, key)
return key
def delete(self, key):
"""
Remove the ``key`` from hash ``name``
:param key:
:return:
"""
self.__conn.hdel(self.name, key)
def inckey(self, key, value):
self.__conn.hincr(self.name, key, value)
def getAll(self):
return self.__conn.hgetall(self.name).keys()
def get_status(self):
"""
Return the number of elements in hash ``name``
:return:
"""
return self.__conn.hsize(self.name)
def changeTable(self, name):
self.name = name
class QueueSSDB(QueueBase.QueueBase):
"""
base class , only provide interface for sub class to implement
"""
def __init__(self, name, host='localhost', port=8888, **kwargs):
QueueBase.QueueBase.__init__(self, name, host, port)
self.__conn = SSDB(connection_pool=BlockingConnectionPool(host=self.host, port=self.port))
#queue
@QueueBase.catch
def put(self, value, *args, **kwargs):
"""
put an item in the back of a queue
:param value:
:param args:
:param kwargs:
:return:
"""
return self.__conn.qpush_back(self.name,
json.dumps(value, ensure_ascii=False).encode('utf-8') if isinstance(value, dict) or isinstance(value, list) else value)
def save(self, value, *args, **kwargs):
"""
put an item in the back of a queue
:param value:
:param args:
:param kwargs:
:return:
"""
return self.__conn.qpush_back(self.name,
json.dumps(value, ensure_ascii = False).encode('utf-8') if isinstance(value,
dict) or isinstance(
value, list) else value)
@QueueBase.catch
def get(self, *args, **kwargs):
"""
get element from the from of queue
:param args:
:param kwargs:
:return:
"""
value = self.__conn.qpop_front(self.name)
return value[0] if value else value
@QueueBase.catch
def getMore(self, *args, **kwargs):
"""
get element from the from of queue
:param args:
:param kwargs:
:return:
"""
value = self.__conn.qpop_front(self.name, **kwargs)
return value
@QueueBase.catch
def size(self, *args, **kwargs):
return self.__conn.qsize(self.name)
@QueueBase.catch
def changeTable(self, name):
"""
change the queue name to operate
:param name:
:return:
"""
self.name = name
@QueueBase.catch
def select_queue(self, name):
"""
change the queue name to operate
:param name:
:return:
"""
self.name = name
@QueueBase.catch
def qclaerQueue(self):
return self.__conn.qclear(self.name)
#KV
@QueueBase.catch
def keySet(self,key,value):
"""
Set the value at key ``name`` to ``value`` .
:param key:
:param value:
:return:
"""
value = json.dumps(value, ensure_ascii = False).encode('utf-8') if isinstance(value,dict) or isinstance(value, list) else value
return self.__conn.set(key,value)
@QueueBase.catch
def keySetx(self,name, value, ttl=-1):
"""
Set the value of key ``name`` to ``value`` that expires in ``ttl``
seconds. ``ttl`` can be represented by an integer or a Python
timedelta object.
:param name:
:param value:
:param ttl:
:return:
"""
return self.__conn.setx(name,value,ttl=ttl)
@QueueBase.catch
def keyTtl(self,key):
"""
Returns the number of seconds until the key ``name`` will expire.
:return:
"""
self.__conn.ttl(key)
@QueueBase.catch
def keyGet(self,key):
"""
Return the value at key ``name``, or ``None`` if the key doesn't exist
:param key:
:return:
"""
return self.__conn.get(key)
@QueueBase.catch
def keyDel(self,key):
"""
Delete the key specified by ``name`` .
:param key:
:return:
"""
return self.__conn.delete(key)
@QueueBase.catch
def keyKeys(self,key_start='',key_end=''):
"""
Return a list of the top ``limit`` keys between ``name_start`` and
``name_end``
:param key_start:
:param key_end:
:return:
"""
return self.__conn.keys(name_start=key_start,name_end=key_end,limit=100000)
@QueueBase.catch
def keyexists(self,key):
"""
:param key:
:return:
"""
return self.__conn.exists(key)
#SET
@QueueBase.catch
def zsetSet(self,field,score = 1):
if field:
if isinstance(field, dict) or isinstance(field, list):
field = json.dumps(field)
field = field if len(field) < 100 else field[:100]
return self.__conn.zset(self.name, field, score)
@QueueBase.catch
def zgetSet(self,key):
return self.__conn.zget(self.name,key)
@QueueBase.catch
def zexistsSet(self,name,field):
return self.__conn.zexists(name,field)
@QueueBase.catch
def zkeysSet(self):
return self.__conn.zkeys(self.name,'','','',limit=100000000)
@QueueBase.catch
def zdelSet(self,key):
return self.__conn.zdel(self.name,key)
@QueueBase.catch
def multi_zgetSet(self,*keys):
return self.__conn.multi_zget(self.name,*keys)
#Hash
@QueueBase.catch
def hgetallHash(self,key):
return self.__conn.hgetall(key)
@QueueBase.catch
def hincrHash(self,name,key):
return self.__conn.hincr(name,key,amount=1)
@QueueBase.catch
def multi_hsetHash(self,name,**mapping):
return self.__conn.multi_hset(name, **mapping)
@QueueBase.catch
def hlistHash(self,start,end):
return self.__conn.hlist(start, end, limit = 10000000)
@QueueBase.catch
def hclearHash(self,key):
return self.__conn.hclear(key)
@QueueBase.catch
def hset(self, key,value):
return self.__conn.hset(self.name, key,json.dumps(value, ensure_ascii = False).encode('utf-8')
if isinstance(value,dict) or isinstance(value, list) else value)
@QueueBase.catch
def hsize(self):
return self.__conn.hsize(self.name)
@QueueBase.catch
def hget(self, key = None):
if key:
return self.__conn.hget(self.name, key)
else:
if self.__conn.hsize(self.name) > 0:
keys=self.__conn.hkeys(self.name,"", "",limit=1)
if keys:
key=keys[0]
v=self.__conn.hget(self.name, key)
self.__conn.hdel(self.name, key)
return v