class IpdownCommand(BaseCommand):
limit = 200 # 用于手机号分页
dir_path = os.path.dirname(os.path.abspath(__file__)) # 获取commands绝对目录
delimiter = os.sep # 目录分隔符
phone_dir = 'phone_ip'
fileUtil = ''
oUserExtend = ''
oUser = ''
def __init__(self):
self.fileUtil = fileUtil()
# 用户扩展表
self.oUserExtend = YiUserExtend()
# 1. 连一亿元库用户表
self.oUser = YiUser()
def runData(self, date_time):
# 开始时间
start = time.time()
# 时间处理
start_time = datetime.strptime(date_time + " 00:00:00",
'%Y-%m-%d %H:%M:%S')
end_time = datetime.strptime(date_time + " 23:59:59",
'%Y-%m-%d %H:%M:%S')
print("开始时间：%s - 结束时间：%s" % (start_time, end_time))
# 2. 查找有多少条用户
total = self.oUser.getMobileCount(start_time, end_time)
if not total:
print("\"%s\" - \"%s\" 间时区间暂无数据！" % (start_time, end_time))
return None
limit = self.limit
pages = math.ceil(total / limit)
cur_page = 0
# 开始时间
start = time.time()
save_path = "%s%s%s%s" % (self.dir_path, self.delimiter,
self.phone_dir, self.delimiter)
# 创建目录
self.fileUtil.mkdir(save_path)
# 保存文件名
filename = "%s%s%s" % (save_path, date_time, '.txt')
# 通讯详单目录
ssdb_path = "%s%s%s" % (save_path, date_time, self.delimiter)
# 创建目录
self.fileUtil.mkdir(ssdb_path)
num = 0
# 判断文件是否存在，存在就删除
if os.path.exists(filename):
os.remove(filename)
num = 0
# 8.循环手机号
while cur_page < pages:
offset = cur_page * limit
# 自增
cur_page += 1
mobile_data = self.oUser.getMobileUser(start_time, end_time,
offset, limit)
user_phone = ''
for c_phone in mobile_data:
num += 1
phone = c_phone.mobile.strip()
user_phone += str(phone) + "\n"
user_id = str(c_phone.user_id).strip()
# 查找用户扩展信息是否存在
ip_str = self.ipHandler(user_id)
if not ip_str:
continue
#保存ip文件
# 判断目录是否存在
ip_filename = "%s%s%s" % (ssdb_path, phone, '.txt')
if os.path.exists(ip_filename):
os.remove(ip_filename)
self.fileUtil.writePhoneText(ip_filename, json.dumps(ip_str))
num += len(ip_str)
#保存手机号
self.fileUtil.writeText(filename, user_phone)
save_num = "%s 条" % str(num)
print(save_num)
# 结束时间
end = time.time()
# 耗时
time_consuming = "耗时：" + str(end - start)
print(time_consuming)
print("done!")
'''
ip处理
'''
def ipHandler(self, user_id):
if not user_id:
return False
# 要找用户ip
ip_user_info = self.oUserExtend.getUserIp(user_id)
if not ip_user_info:
return False
ip_str = ip_user_info.reg_ip.strip()
if not ip_str:
return False
# 通过ip查找手机号
# print(ip_str)
ip_cpunt = self.oUserExtend.getIpCount(ip_str)
limit = self.limit
pages = math.ceil(ip_cpunt / limit)
cur_page = 0
user_phone = []
while cur_page < pages:
offset = cur_page * limit
# 自增
cur_page += 1
# 获取同一个ip的所有信息
get_ip_data = self.oUserExtend.getIpData(ip_str, offset, limit)
if not get_ip_data:
continue
#获取用户user_id
ip_tuple = []
for ip_info in get_ip_data:
ip_tuple.append(str(ip_info.user_id))
# 去掉重复的user_id
if user_id in ip_tuple:
ip_tuple.remove(user_id)
# 判断是否为空，为空跳过
if not ip_tuple:
continue
ip_tuple = tuple(ip_tuple)
#判断是否需要in
if len(ip_tuple) > 1:
user_data = self.oUser.getMobileByUserId(ip_tuple)
user_phone.extend(user_data)
if (len(ip_tuple)) == 1:
user_id = ip_tuple[0]
user_info = self.oUser.getByUserId(user_id)
try:
if user_info.mobile:
user_phone.append(user_info.mobile)
except AttributeError as f:
print(f)
continue
return user_phone
def runData(self, start_time, end_time):
# 时间处理
start_time = datetime.strptime(start_time + " 00:00:00", '%Y-%m-%d %H:%M:%S')
end_time = datetime.strptime(end_time + " 23:59:59", '%Y-%m-%d %H:%M:%S')
print("开始时间：%s - 结束时间：%s" % (start_time, end_time))
# 1. 连一亿元库用户表
oUser = YiUser()
# 2. 查找有多少条用户
total = oUser.getMobileCount(start_time, end_time)
if not total:
print("\"%s\" - \"%s\" 时间区间暂无数据！" % (start_time, end_time))
return None
limit = self.limit
pages = math.ceil(total / limit);
cur_page = 0
# 3. 连接图数据库
client = self.Oorientdb.connectDb()
if not client:
return False
# 4. 打开数据库
open_database = self.Oorientdb.openDatabas(client)
if not open_database:
return False
# 5.查看类是否存在
self.Oorientdb.createClass(client, self.class_name)
# 6.创建属性
self.Oorientdb.createProperty(client, self.class_name, self.property)
# 7.创建索引
self.Oorientdb.createIndex(client, self.class_name, self.property)
# 创建边缘
self.Oorientdb.createEdge(client, self.edge_name)
num = 0
insert_num = 0
# 开始时间
start = time.time()
# 8.循环手机号
while cur_page < pages:
offset = cur_page * limit
# 自增
cur_page += 1
mobile_data = oUser.getMobileUser(start_time, end_time, offset, limit)
for user_info in mobile_data:
phone = user_info.mobile.strip()
# 9.从ssdb中获取数数据
phone_detail = self.ssdb_resources.get(phone)
# phone_detail = '{"phoneArr":["15125954590","073188040149","15125887806","18719499607","18287885870","13764022964"],"create_time":"2018-03-15 09:52;01","modify_time":"2018-06-24 12:30:06"}'
message_list = []
if phone_detail is not None:
message_list = json.loads(phone_detail)
message_list = list(set(message_list['phoneArr']))
# 去掉当前手机号
if phone in message_list:
message_list.remove(phone)
num += len(message_list)
# ssdb分页
ssdb_limit = self.ssdb_limit;
ssdb_cur_page = 0
while ssdb_cur_page < len(message_list):
ssdb_offset = ssdb_cur_page + ssdb_limit
print("%s---%d---%d" % (phone, ssdb_cur_page, ssdb_offset))
page_message_list = message_list[ssdb_cur_page:ssdb_offset]
ssdb_cur_page += ssdb_limit
# 将当前手机号加入到查询列表中
page_message_list.append(phone)
# 批量查找
get_mobile_data = self.Oorientdb.getMobileData(client, self.class_name, page_message_list)
source_rid = ''
target_list = ""
for m in get_mobile_data:
if m.phone == phone:
source_rid = m._OrientRecord__rid
continue
target_list += m._OrientRecord__rid + ","
target_list = target_list.strip("\,")
# 如果来源为空或目标为空就跳过
if not source_rid or not target_list:
continue
#查找边缘是否存在
get_edge_data = self.Oorientdb.getEdgeData(client, self.edge_name, source_rid)
edge_list = ""
for edge_data in get_edge_data:
#去掉当前手机号
if source_rid == edge_data._OrientRecord__rid:
continue
edge_list += edge_data._OrientRecord__rid + ","
edge_list = edge_list.strip("\,")
# 去掉重复的rid
target_rid = list(set(target_list).difference(set(edge_list)))
if not target_rid:
continue
# 创建成功累加
insert_num += len(target_rid)
#批查插入
insert_edge = self.Oorientdb.createEdgeMore(client, self.edge_name,source_rid, target_rid)
# 结束时间
end = time.time()
# 耗时
time_consuming = "耗时：" + str(end - start)
print(time_consuming)
print("扫描：%s 条" % str(num))
print("创建：%s 条" % str(insert_num))
print("done!")
def runData(self, start_time = None , end_time = None):
# self.example();
# return None
#=======================================
# 开始时间
start = time.time()
#1. 判断开始和结束时间，如果不存在就返回当前时间
last_day = datetime.now() + timedelta(days=-1)
if start_time is None:
start_time = last_day.strftime('%Y-%m-%d 00:00:00')
if end_time is None:
end_time = last_day.strftime('%Y-%m-%d 23:59:59')
# =======================================
#2. 连接ssdb
ssdb = SsdbObject(False)
ssdb_resources = ssdb.ssdbConnection()
# print(ssdb_resources)
# =======================================
#3. 连一亿元库用户表
oUser = YiUser()
# 5.连接orientdb数据库
client = self.connectDb()
# 6.打开数据库
open_databases = self.openData(client)
# 7.创建类和边缘
class_name = self.class_name # 类名
dege_name = self.dege_name # 边缘名
class_info = self.createClass(client, class_name) # 创建类
dege_info = self.createClass(client, dege_name, False) # 创建边缘
index_name = self.createIndex(client, class_name, "phone") # 创建索引
# =======================================
# 5.查找有多少条用户
total = oUser.getMobileCount(start_time, end_time)
# 6.分页处理
limit = self.limit
pages = math.ceil(total / limit);
cur_page = 0;
num = 0
while cur_page < pages:
offset = cur_page * limit
mobile_data = oUser.getMobileUser(start_time, end_time, offset, limit)
for data in mobile_data:
phone = data.mobile.strip()
# 查找关系数据
get_data = self.getData(client, class_name, phone)
# 创建顶点
phone_info = self.createVertex(client, class_name, phone)
source = phone_info[0].rid
#========================================
# 9.ssdb找到用户通讯录
if not phone:
continue
message_str = ssdb_resources.get(phone)
message_list = []
if message_str is not None:
message_list = list(set(json.loads(message_str)))
# 去掉当前手机号
if phone in message_list:
message_list.remove(phone)
#ssdb分页
ssdb_limit = self.ssdb_limit;
ssdb_cur_page = 0
while ssdb_cur_page < len(message_list):
ssdb_offset = ssdb_cur_page + ssdb_limit
print("%s---%d---%d" % (phone, ssdb_cur_page, ssdb_offset))
page_message_list = message_list[ssdb_cur_page:ssdb_offset]
#print(page_message_list)
#return None
ssdb_cur_page += ssdb_limit
# print(page_message_list)
# 插入数据
insert_bool = self.insertValues(client, class_name, page_message_list)
# if not insert_bool:
# pass
# continue
# 获取插入数据
get_mobile = self.getMobileData(client, class_name, page_message_list)
target_list = ""
for target in get_mobile:
target_list += target._OrientRecord__rid + ","
target_list = target_list.strip("\,")
if not target_list:
continue
# 创建边缘
edge_bool = self.createEdgeMore(client, dege_name, source, target_list)
num += 1
# # print(page_message_list)
# # 插入数据
# insert_bool = self.insertValues(client, class_name, page_message_list)
# if not insert_bool:
# continue
# target_list = ""
# for target in insert_bool:
# target_list += target._OrientRecord__rid + ","
# target_list = target_list.strip("\,")
# if not target_list:
# continue
# #创建边缘
# edge_bool = self.createEdgeMore(client, dege_name, source, target_list)
# num += 1
# # 休息1秒中
# time.sleep(1)
return None
#自增
cur_page += 1
print(num)
# 结束时间
end = time.time()
# 耗时
time_consuming = "耗时" + str(end - start)
print(time_consuming)
print("done!")
def runData(self, date_time):
# 开始时间
start = time.time()
# 连一亿元库用户表
oUser = YiUser()
# 时间
check_data = self.checkData(date_time)
if check_data == False:
print("时间格式错误!")
return None
# =======================================
# 2. 连接ssdb
ssdb = SsdbObject(False)
ssdb_resources = ssdb.ssdbConnection()
# print(ssdb_resources)
# =======================================
# 3. 连一亿元库用户表
oUser = YiUser()
# 5.连接orientdb数据库
client = self.connectDb()
# 6.打开数据库
open_databases = self.openData(client)
# 7.创建类和边缘
class_name = self.class_name # 类名
dege_name = self.dege_name # 边缘名
class_info = self.createClass(client, class_name) # 创建类
dege_info = self.createClass(client, dege_name, False) # 创建边缘
index_name = self.createIndex(client, class_name, "phone") # 创建索引
# =======================================
# 5.查找有多少条用户
total = oUser.getMobileCount(self.start_time, self.end_time)
# 6.分页处理
limit = self.limit
pages = math.ceil(total / limit)
cur_page = 0
num = 0
while cur_page < pages:
offset = cur_page * limit
mobile_data = oUser.getMobileUser(self.start_time, self.end_time,
offset, limit)
for data in mobile_data:
phone = data.mobile.strip()
# 查找关系数据
get_data = self.getData(client, class_name, phone)
# 创建顶点
phone_info = self.createVertex(client, class_name, phone)
source = phone_info[0].rid
# ========================================
# sdb找到用户通讯录
if not phone:
continue
message_str = ssdb_resources.get(phone)
message_list = []
if message_str is not None:
message_list = list(set(json.loads(message_str)))
# 去掉当前手机号
if phone in message_list:
message_list.remove(phone)
# ssdb分页
ssdb_limit = self.ssdb_limit
ssdb_cur_page = 0
while ssdb_cur_page < len(message_list):
ssdb_offset = ssdb_cur_page + ssdb_limit
print("%s---%d---%d" % (phone, ssdb_cur_page, ssdb_offset))
page_message_list = message_list[ssdb_cur_page:ssdb_offset]
ssdb_cur_page += ssdb_limit
# 插入数据
insert_bool = self.insertValues(client, class_name,
page_message_list)
# 获取插入数据
get_mobile = self.getMobileData(client, class_name,
page_message_list)
if not get_mobile:
continue
#过滤掉重复的边缘
target_list = self.filterEdge(client, dege_name, source,
get_mobile)
if not target_list:
continue
# # 创建边缘
edge_bool = self.createEdgeMore(client, dege_name, source,
target_list)
num += len(edge_bool)
# 自增
cur_page += 1
print("共创建 %d 条边缘" % num)
# 结束时间
end = time.time()
# 耗时
time_consuming = "耗时" + str(end - start)
print(time_consuming)
print("done!")
class MaillistCommand(BaseCommand):
def __init__(self):
config = OrientdbConfig().getConfig()
self.host = config['http_host']
self.port = config['http_port']
self.username = config['user']
self.passwd = config['password']
self.db_name = 'graph_data'
#self.db_name = 'graphdata1'
self.limit = 500 #用于手机号分页
self.oUser = YiUser()
"""运行数据"""
def runData(self, start_time=None, end_time=None):
# 1. 判断开始和结束时间，如果不存在就返回当前时间
last_day = datetime.now() + timedelta(days=-1)
if start_time is None:
start_time = last_day.strftime('%Y-%m-%d 00:00:00')
if end_time is None:
end_time = last_day.strftime('%Y-%m-%d 23:59:59')
# =======================================
#连接ssdb
ssdb = SsdbObject(False)
ssdb_resources = ssdb.ssdbConnection()
#开始时间
start = time.time()
# 连接一亿元用户表
#查找有多少条用户
total = self.oUser.getMobileCount(start_time, end_time)
limit = self.limit
pages = math.ceil(total / limit)
cur_page = 0
num = 0
#目录
path_name = "./commands/mobile/"
path_a = self.mkdir(path_name)
#打开文件
mobile_filename = path_name + start_time + ".txt"
fo = open(mobile_filename, "w+")
while cur_page < pages:
#分页到出数据
offset = cur_page * limit
mobile_data = self.oUser.getMobileUser(start_time, end_time,
offset, limit)
mobile_str = ''
for mobile in mobile_data:
#单条词ssdb数据
phone = mobile.mobile.strip()
print("user_mobile=%s" % phone)
if not phone:
continue
ssdb_message_str = ssdb_resources.get(phone)
if ssdb_message_str:
ssdb_path_name = path_name + start_time + "/"
path_s = self.mkdir(ssdb_path_name)
ssdb_filename = ssdb_path_name + phone + ".json"
fo_ssdb = open(ssdb_filename, "wb+")
fo_ssdb.write(ssdb_message_str)
mobile_str += phone + "\n"
fo.write(mobile_str)
num += 1
# 自增
cur_page += 1
print(num)
end = time.time()
#耗时
time_consuming = "耗时" + str(end - start)
print(time_consuming)
#fo.write(time_consuming)
fo.close()
print("done!")
#end_time = time.time()
"""
创建目录
"""
def mkdir(self, path):
# 去除首位空格
path = path.strip()
# 去除尾部 \ 符号
path = path.rstrip("\\")
# 判断路径是否存在
# 存在 True
# 不存在 False
isExists = os.path.exists(path)
# 判断结果
if not isExists:
# 如果不存在则创建目录
# 创建目录操作函数
os.makedirs(path)
print(path + ' 创建成功')
return True
else:
# 如果目录存在则不创建，并提示目录已存在
print(path + ' 目录已存在')
return False
class ColonymoreCommand(BaseCommand):
# 数据库类对象
oUser = ''
def __init__(self):
self.oUser = YiUser()
#
def runData(self, start_time, end_time):
#时间处理
start_time = datetime.strptime(start_time + " 00:00:00",
'%Y-%m-%d %H:%M:%S')
end_time = datetime.strptime(end_time + " 23:59:59",
'%Y-%m-%d %H:%M:%S')
#数据库操作
self.userOperation(start_time, end_time)
def userOperation(self, start_time, end_time):
# 查找有多少条用户
total = self.oUser.getMobileCount(start_time, end_time)
print(total)
'''
集群图数据库配置
'''
def orientdbConfig(self):
orientdb_config = {
"http_host": "47.96.99.175",
"http_port": "2424",
"user": "******",
"password": "******",
"db_name": "test", #数据库名
"class_name": "class_phone", #类名
"ip_edge_name": "ip_relation_ship", #ip边缘名
"specifications_edge_name":
"specifications_relation_ship", # 详单边缘名
"maillist_edge_name": "maillist_relation_ship", # 通讯录边缘名
}
return orientdb_config
'''
通讯录ssdb配置
'''
def maillistSsdbConfig(self):
ip_data = {
"master": "47.93.121.71:8888",
"slave": {
"47.93.121.71:8888": 4,
"47.93.121.71:8888": 6
},
}
return ip_data
'''
通话详单ssdb配置
'''
def specificationsConfig(self):
ip_data = {
"master": "47.93.121.71:8888",
"slave": {
"47.93.121.71:8888": 4,
"47.93.121.71:8888": 6
},
}
return ip_data
'''
clickhouse配置
'''
def clickhouseConfig(self):
click_config = {
"http": "47.93.121.71",
"port": "9001",
"user": "******",
"pwd": "6lYaUiFi",
"database": "",
}
return click_config
'''
clickhouse连接
'''
def clickhouseConnection(self):
config = self.clickhouseConfig()
# 连接数据库
client = Client(config["http"], config["port"], config["database"],
config["user"], config["pwd"])
a = client.execute("show tables")
print(a)
def runData(self, start_time, end_time):
# 时间处理
start_time = datetime.strptime(start_time + " 00:00:00", '%Y-%m-%d %H:%M:%S')
end_time = datetime.strptime(end_time + " 23:59:59", '%Y-%m-%d %H:%M:%S')
print("开始时间：%s - 结束时间：%s" % (start_time, end_time))
# 1. 连一亿元库用户表
oUser = YiUser()
# 2. 查找有多少条用户
total = oUser.getMobileCount(start_time, end_time)
if not total:
print("\"%s\" - \"%s\" 时间区间暂无数据！" % (start_time, end_time))
return None
limit = self.limit
pages = math.ceil(total / limit);
cur_page = 0
# 3. 连接图数据库
client = self.Oorientdb.connectDb()
if not client:
return False
# 4. 打开数据库
open_database = self.Oorientdb.openDatabas(client)
if not open_database:
return False
# 5.查看类是否存在
self.Oorientdb.createClass(client, self.class_name)
# 6.创建属性
self.Oorientdb.createProperty(client, self.class_name, self.property)
# 7.创建索引
self.Oorientdb.createIndex(client, self.class_name, self.property)
num = 0
insert_num = 0
# 开始时间
start = time.time()
# 8.循环手机号
while cur_page < pages:
offset = cur_page * limit
# 自增
cur_page += 1
mobile_data = oUser.getMobileUser(start_time, end_time, offset, limit)
for user_info in mobile_data:
phone = user_info.mobile.strip()
# 9.从ssdb中获取数数据
phone_detail = self.ssdb_resources.get(phone)
# phone_detail = '{"phoneArr":["15125954590","073188040149","15125887806","18719499607","18287885870","13764022964"],"create_time":"2018-03-15 09:52;01","modify_time":"2018-06-24 12:30:06"}'
message_list = []
if phone_detail is not None:
message_list = json.loads(phone_detail)
message_list = list(set(message_list['phoneArr']))
# 去掉当前手机号
if phone in message_list:
message_list.remove(phone)
# 加入当前手机号到通讯详单中
message_list.append(phone)
num += len(message_list)
# ssdb分页
ssdb_limit = self.ssdb_limit;
ssdb_cur_page = 0
while ssdb_cur_page < len(message_list):
ssdb_offset = ssdb_cur_page + ssdb_limit
print("%s---%d---%d" % (phone, ssdb_cur_page, ssdb_offset))
page_message_list = message_list[ssdb_cur_page:ssdb_offset]
ssdb_cur_page += ssdb_limit
# 批量获取数据
get_batch_data = self.Oorientdb.getBatchData(client, self.class_name, page_message_list)
# 去掉重复的数据
mobile_data = list(set(page_message_list).difference(set(get_batch_data)))
#mobile_data = get_batch_data
# 格式数据
if not mobile_data:
continue
phone_tuple = ''
for mobile in mobile_data:
phone_tuple += "('" + mobile + "'),"
phone_tuple = phone_tuple.strip("\,")
# 批量插入
insert_batch_data = self.Oorientdb.insertBatchData(client, self.class_name, phone_tuple)
if not insert_batch_data:
continue
insert_num += len(mobile_data)
# 结束时间
end = time.time()
# 耗时
time_consuming = "耗时：" + str(end - start)
print(time_consuming)
print("扫描：%s 条" % str(num))
print("插入：%s 条" % str(insert_num))
print("done!")
def runData(self, date_time):
# 开始时间
start = time.time()
# 时间处理
start_time = datetime.strptime(date_time + " 00:00:00",
'%Y-%m-%d %H:%M:%S')
end_time = datetime.strptime(date_time + " 23:59:59",
'%Y-%m-%d %H:%M:%S')
print("开始时间：%s - 结束时间：%s" % (start_time, end_time))
# 1. 连一亿元库用户表
oUser = YiUser()
# 2. 查找有多少条用户
total = oUser.getMobileCount(start_time, end_time)
if not total:
print("\"%s\" - \"%s\" 间时区间暂无数据！" % (start_time, end_time))
return None
limit = self.limit
pages = math.ceil(total / limit)
cur_page = 0
# 开始时间
start = time.time()
save_path = "%s%s%s%s" % (self.dir_path, self.delimiter,
self.phone_dir, self.delimiter)
#创建目录
self.mkdir(save_path)
#保存文件名
filename = "%s%s%s" % (save_path, date_time, '.txt')
# 通讯详单目录
ssdb_path = "%s%s%s" % (save_path, date_time, self.delimiter)
# 创建目录
self.mkdir(ssdb_path)
num = 0
# 判断文件是否存在，存在就删除
if os.path.exists(filename):
os.remove(filename)
# 8.循环手机号
while cur_page < pages:
offset = cur_page * limit
# 自增
cur_page += 1
mobile_data = oUser.getMobileUser(start_time, end_time, offset,
limit)
user_phone = ''
for c_phone in mobile_data:
num += 1
#查找ssdb上的数据
phone = c_phone.mobile.strip()
user_phone += str(phone) + "\n"
# 判断目录是否存在
ssdb_filename = "%s%s%s" % (ssdb_path, phone, '.txt')
phone_detail = self.ssdb_resources.get(phone)
#phone_detail = '{"phoneArr":["15125954590","073188040149","15125887806","18719499607","18287885870","13764022964"],"create_time":"2018-03-15 09:52;01","modify_time":"2018-06-24 12:30:06"}'
if not phone_detail:
continue
print(type(phone_detail))
if os.path.exists(ssdb_filename):
os.remove(ssdb_filename)
# 记录到文件中
ssdb_save = self.writeDetailText(ssdb_filename, phone_detail)
if ssdb_save:
print("保存 \"%s\" 详单成功" % phone)
else:
print("保存 \"%s\" 详单失败" % phone)
if not user_phone:
continue
#记录到文件中
user_phone_save = self.writeText(filename, user_phone)
if user_phone_save:
print("%s用户手机号保存成功" % user_phone)
else:
print("%s用户手机号保存失败" % user_phone)
save_num = "%s 条" % str(num)
print(save_num)
# 结束时间
end = time.time()
# 耗时
time_consuming = "耗时：" + str(end - start)
print(time_consuming)
print("done!")
class ColonylistCommand(BaseCommand):
# 图数据库连接句柄
orientdb_client = ''
# 图数据库配置文件
orientdb_config = ''
# 数据库类对象
oUser = ''
# 连接ssdb
ssdb_resources = ""
# 获取commands绝对目录
dir_path = os.path.dirname(os.path.abspath(__file__))
# 目录分隔符
delimiter = os.sep
phone_dir = 'mobile'
key_suffix = '_orientdb_10'
# 图数据库名
#db_name = 'graph_phone'
db_name = 'test'
# ssdb通讯录分页，用于限制一次可以插入多少条
ssdb_limit = 500
limit = 200
# 类名
class_name = "class_phone_test"
# 边缘名
edge_name = "relation_ship_test"
orientdb_config = {
"http_host": "47.96.99.175",
"http_port": "2424",
"user": "******",
"password": "******",
"database": "test",
}
def __init__(self):
#self.orientdb_config = OrientdbConfig().getConfig()
self.oUser = YiUser()
self.orientdb_client = self.connectDb()
# 连接ssdb
ssdb = SsdbObject(False)
self.ssdb_resources = ssdb.ssdbConnection()
def runData(self, start_time, end_time):
# 查找有多少条用户
total = self.oUser.getMobileCount(start_time, end_time)
if not total:
print("没有用户信息")
return False
# 图数据库
orien_oper = self.orientdbOperation()
if not orien_oper:
print("图数据库操作失败")
return False
# 分页处理
limit = self.limit
limit = 2
pages = math.ceil(total / limit)
cur_page = 0
num = 0
# 开始时间
start = time.time()
while cur_page < pages:
offset = cur_page * limit
# 自增
cur_page += 1
# 取出用户信息
mobile_data = self.oUser.getMobileUser(start_time, end_time,
offset, limit)
for user_info in mobile_data:
phone = user_info.mobile.strip()
mail_list_num = self.importOperation(phone)
num += mail_list_num
# 结束时间
end = time.time()
# 耗时
time_consuming = str(num) + " 条数据耗时：" + str(end - start)
print(time_consuming)
def importOperation(self, phone):
# 判断手机号是否存在在缓存中
key = phone + self.key_suffix
phone_rid = self.ssdb_resources.get(key)
if not phone_rid:
try:
p_rid = self.insertOne(phone)
self.ssdb_resources.set(key, p_rid)
phone_rid = self.ssdb_resources.get(key)
except pyorient.exceptions.PyOrientORecordDuplicatedException as f:
print(f)
vertext_info = self.getVertext(phone)
v_rid = "#" + vertext_info[0].rid._OrientRecordLink__link
self.ssdb_resources.set(key, v_rid)
phone_rid = self.ssdb_resources.get(key)
#查找通讯录
message_str = self.ssdb_resources.get(phone)
message_list = []
if message_str is not None:
message_list = list(set(json.loads(message_str)))
# 去掉当前手机号
if phone in message_list:
message_list.remove(phone)
num = len(message_list) + 1
#====================================================
# ssdb分页
ssdb_limit = self.ssdb_limit
ssdb_cur_page = 0
while ssdb_cur_page < len(message_list):
ssdb_offset = ssdb_cur_page + ssdb_limit
print("%s---%d---%d" % (phone, ssdb_cur_page, ssdb_offset))
page_message_list = message_list[ssdb_cur_page:ssdb_offset]
ssdb_cur_page += ssdb_limit
# 插入数据
try:
insert_bool = self.insertValues(page_message_list)
except pyorient.exceptions.PyOrientORecordDuplicatedException as f:
print(f)
# 获取rid
target_list = ""
for target_phone in page_message_list:
t_key = target_phone + self.key_suffix
t_rid = self.ssdb_resources.get(t_key)
if not t_rid:
continue
target_list += str(t_rid, encoding="utf-8") + ","
target_list = target_list.strip("\,")
#print(phone_rid)
phone_rid = str(phone_rid, encoding="utf-8")
# 创建边缘
try:
edge_bool = self.createEdgeMore(phone_rid, target_list)
except pyorient.exceptions.PyOrientCommandException as f:
print(f)
return num
'''
图数据库操作（类，边缘，数据库，索引）
'''
def orientdbOperation(self):
# 1.判断本地数据库是否存在
is_database = self.orientdb_client.db_exists(
self.db_name, pyorient.STORAGE_TYPE_PLOCAL)
if not is_database:
print("\"%s\" 数据库不存在" % self.db_name)
return False
# 2.打开数据库
try:
open_databases = self.orientdb_client.db_open(
self.db_name, self.orientdb_config['user'],
self.orientdb_config['password'])
except pyorient.exceptions.PyOrientDatabaseException as e:
print(e)
return False
# 3.创建类
try:
create_class_sql = "CREATE CLASS %s extends %s" % (self.class_name,
"V")
create_class = self.orientdb_client.command(create_class_sql)
except pyorient.exceptions.PyOrientSchemaException as p:
print("创建类：" + "=" * 50)
print(p)
# 4.创建类的属性
try:
class_property_sql = "CREATE PROPERTY %s.phone STRING" % self.class_name
#print(class_property_sql)
create_property = self.orientdb_client.command(
class_property_sql) # 手机号
#print(create_property)
except pyorient.exceptions.PyOrientCommandException as p:
print("创建类属性:" + "=" * 50)
print(p)
# 5.创建边缘
try:
edge_sql = "CREATE CLASS %s extends %s" % (self.edge_name, "E")
#print(edge_sql)
create_edge = self.orientdb_client.command(edge_sql)
#print(create_edge)
except pyorient.exceptions.PyOrientSchemaException as f:
print("创建边缘:" + "=" * 50)
print(f)
# 6.创建索引
try:
index_sql = "create index %s ON %s(%s) UNIQUE " % (
self.class_name + "_phone", self.class_name, "phone")
create_edge = self.orientdb_client.command(index_sql)
print(create_edge)
except pyorient.exceptions.PyOrientIndexException as f:
print("创建索引:" + "=" * 50)
print(f)
print("创建数据库相关结束" + "=" * 40)
return True
'''
连接图数据库
'''
def connectDb(self):
client = pyorient.OrientDB(self.orientdb_config['http_host'],
int(self.orientdb_config['http_port']))
session_id = client.connect(self.orientdb_config['user'],
self.orientdb_config['password'])
return client
'''
单条插入
'''
def insertOne(self, phone):
insert_str = "insert into %s (phone) values(%s)" % (self.class_name,
phone)
insert_info = self.orientdb_client.command(insert_str)
return insert_info[0]._OrientRecord__rid
'''
插入数据
'''
def insertValues(self, message_list):
ssdb_mobile = []
for mobile in message_list:
m_key = mobile + self.key_suffix
if self.ssdb_resources.get(m_key):
continue
ssdb_mobile.append(mobile)
if len(ssdb_mobile) < 1:
return False
# 表数过滤
diff_mobile = self.getMobileAll(ssdb_mobile)
# 判断是否存在
if len(diff_mobile) < 1:
return False
str = ''
for mobile in diff_mobile:
str += "('" + mobile + "'),"
str = str.strip("\,")
# 批量插入
insert_str = "insert into %s (phone) values %s" % (self.class_name,
str)
insert_data = self.orientdb_client.command(insert_str)
for insert_object in insert_data:
p_rid = insert_object._OrientRecord__rid
key = insert_object.phone + self.key_suffix
if not self.ssdb_resources.get(key):
self.ssdb_resources.set(key, p_rid)
return True
'''
创建边缘批量
'''
def createEdgeMore(self, source, target):
edget_sql = "CREATE EDGE %s from %s TO [%s] " % (self.edge_name,
source, target)
ret = self.orientdb_client.command(edget_sql)
#print(ret)
return ret
'''
查找顶点
'''
def getVertext(self, phone):
return self.orientdb_client.command(
'select @rid from %s where phone="%s" ' % (self.class_name, phone))
'''
批量获取数据
'''
def getMobileAll(self, v_values):
# 批量获取数据
get_sql = "select from %s where phone in %s" % (self.class_name,
v_values)
get_data = self.orientdb_client.command(get_sql)
mobile_data = []
for mobile in get_data:
m_key = mobile.phone + self.key_suffix
#print(m_key)
self.ssdb_resources.set(m_key, mobile._OrientRecord__rid)
mobile_data.append(mobile.phone)
#===============================
#v_values中有而mobile_data中没有的
mobile_data = list(set(v_values).difference(set(mobile_data)))
return mobile_data