python hdfs dataframe python读取hdfs并返回dataframe教程

from hdfs import Client
import pandas as pd
 
HDFSHOST = "http://xxx:50070"
FILENAME = "/tmp/preprocess/part-00000" #hdfs文件路径
COLUMNNAMES = [xx']
 
def readHDFS():
'''
读取hdfs文件
 
Returns：
df:dataframe hdfs数据
'''
client = Client(HDFSHOST)
# 目前读取hdfs文件采用方式：
# 1. 先从hdfs读取二进制数据流文件
# 2. 将二进制文件另存为.csv
# 3. 使用pandas读取csv文件
with client.read(FILENAME) as fs:
content = fs.read()
s = str(content, 'utf-8')
file = open("data/tmp/data.csv", "w")
file.write(s)
df = pd.read_csv("data/tmp/data.csv", names=COLUMNNAMES)
return df

from pyhdfs import HdfsClient
client=HdfsClient(hosts='ghym:50070')#hdfs地址
res=client.open('/sy.txt')#hdfs文件路径,根目录/
for r in res:
  line=str(r,encoding='utf8')#open后是二进制,str()转换为字符串并转码
  print(line)

from pyhdfs import HdfsClient
client=HdfsClient(hosts='ghym:50070',user_name='hadoop')#只有hadoop用户拥有写权限
str='hello world'
client.create('/py.txt',str)#创建新文件并写入字符串

from pyhdfs import HdfsClient
client = HdfsClient(hosts='ghym:50070', user_name='hadoop')
client.copy_from_local('d:/pydemo.txt', '/pydemo')#本地文件绝对路径,HDFS目录必须不存在

# pyhdfs读取文本文件,分隔符为逗号,
from pyhdfs import HdfsClient
client = HdfsClient(hosts='ghym:50070', user_name='hadoop')
inputfile=client.open('/int.txt')
# pandas调用读取方法read_table
import pandas as pd
df=pd.read_table(inputfile,encoding='gbk',sep=',')#参数为源文件,编码,分隔符
# 数据集to_csv方法转换为csv
df.to_csv('demo.csv',encoding='gbk',index=None)#参数为目标文件,编码,是否要索引