python读取kafka,输出到Vertica数据库

# 主测试
# https://docs.python.org/2/library/json.html
import sys
import json
import vertica_python
import time
import os
from pykafka import KafkaClient  # 导入的vertica_python和pykafka包需要pip install安装

# 显示当前时间
print('开始时间', time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
print(sys.getdefaultencoding())
client = KafkaClient(hosts="192.168.1.1:9092")  # 填写kafka地址和端口,一般是9092端口
# client.topics  # 查看所有topic
topic = client.topics[b'topic']  # 选择一个topic
consumer = topic.get_simple_consumer(consumer_timeout_ms=2000, auto_commit_enable=1)  # 等待5秒无新数据,退出
data_group = []
conn_info = {'host': '192.168.1.1', 'port': 1, 'user': 'a', 'password': 'b',
             'database': 'c', 'read_timeout': 600, 'unicode_error': 'strict', 'ssl': False}  # 填写数据库连接信息
# simple connection, with manual close
connection = vertica_python.connect(**conn_info)
cur = connection.cursor()
a_error_count = 0
a_success_count = 0
path_os = os.path.abspath('offset.txt') # 将数据偏移量offset写入文件
f1 = open(path_os, 'r', encoding='utf8')
a_offset_start = int(f1.readline())  # 从a_offset_start开始读数据
print(a_offset_start)
# a_offset_start = 3000 # 可以手工指定从哪里开始读取数据,排错用
f1.close()
for message in consumer:  # 循环0
    if message is not None and message.offset > a_offset_start:
        try:
            a = message.value.decode('UTF-8')
            data_group.append(json.loads(a))
            c = message.offset
            for item in data_group:
                str1 = "insert into 表名(列名) values "+ "('" + str(c)  # 将offset值也写入数据库 
          + "'," + "'%s','%s'); " % ( item['列名1'], item['列名2']) print(str1) cur.execute(str1) connection.commit() a_success_count += 1 data_group.pop() except: print('error_message') a_error_count += 1 continue c1 = message.offset f = open(path_os, 'w+' , encoding='utf8') f.truncate() f.write(str(c1)) f.write(' ' + '开始时间=' + time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time()))) f.write(' ' + 'a_success_count=' + str(a_success_count)) f.write(' ' + 'a_error_count=' + str(a_error_count)) f.close()
原文地址:https://www.cnblogs.com/castlevania/p/6428213.html