python hbase读取数据发送kafka的方法

时间:2021-09-15 17:07:14

本例子实现从hbase获取数据,并发送kafka

使用

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
#!/usr/bin/env python
#coding=utf-8
 
import sys
import time
import json
 
sys.path.append('/usr/local/lib/python3.5/site-packages')
from thrift import Thrift
from thrift.transport import TSocket
from thrift.transport import TTransport
from thrift.protocol import TBinaryProtocol
from hbase1 import Hbase #调用hbase thrif1
from hbase1.ttypes import *
from kafka import KafkaConsumer
from kafka import KafkaProducer
from kafka.errors import KafkaError
import unittest
 
class HbaseOpreator:
 def __init__(self,host,port,table='test'):
  self.tableName=table
  self.transport=TTransport.TBufferedTransport(TSocket.TSocket(host,port))
  self.protocol=TBinaryProtocol.TBinaryProtocol(self.transport)
  self.client=Hbase.Client(self.protocol)
  self.transport.open()
 
 def __del__(self):
  self.transport.close()
 
 
 def scanTablefilter(self,table,*args):
  d=dict()
  L=[]
  try:
   tableName=table
   # scan = Hbase.TScan(startRow, stopRow)
   scan=TScan()
   #主键首字母123
   # filter = "PrefixFilter('123_')"
   # filter = "RowFilter(=,'regexstring:.aaa')"
   #过滤条件,当前为 statis_date 字段,值为20170223
   # fitler = "SingleColumnValueFilter(tableName,'f','statis_date','20170223')"
   # filter="SingleColumnValueFilter('f','statis_date',=,'binary:20170223') AND SingleColumnValueFilter('f','name',=,'binary:LXS')"
   filter="SingleColumnValueFilter('info','name',=,'binary:lilei') OR SingleColumnValueFilter('info','name',=,'binary:lily')"
   scan.filterString=filter
   id=self.client.scannerOpenWithScan(tableName,scan,None)
   result=self.client.scannerGet(id)
   # result=self.client.scannerGetList(id,100)
   while result:
    for r in result:
     key=r.row
     name=r.columns.get('info:name').value
     age=r.columns.get('info:age').value
     phone=r.columns.get('info:phone').value
     d['key']=key
     d['name']=name
     d['age']=age
     d['phone']=phone
     # encode_result_json=json.dumps(d).encode(encoding="utf-8")
     # print(encode_result_json)
     L.append(d)        
    result=self.client.scannerGet(id)   
   return json.dumps(L).encode(encoding="utf-8"
  finally:
   # self.client.scannerClose(scan)
   print("scan finish")
 
def sendKfafkaProduct(data):
 # self.host_port='localhost:9092'
 producer = KafkaProducer(bootstrap_servers=['localhost:9092'])
 for d in data:
  producer.send('test', key=b'lxs', value=d)
  time.sleep(5)
  print(d)
 
 while True:
  producer.send('test', key=b'lxs', value=data)
  time.sleep(5)
  print(data)
 
if __name__== '__main__':
 # unittest.main()
 
 B=HbaseOpreator('10.27.1.138',9090)
 value=B.scanTablefilter('ns_lbi:test_hbase_student')
 print(value)
 #sendKfafkaProduct(value)

以上这篇python hbase读取数据发送kafka的方法就是小编分享给大家的全部内容了,希望能给大家一个参考,也希望大家多多支持服务器之家。

原文链接:https://blog.csdn.net/meiguopai1/article/details/70175069