因在公司业务上遇到一个需求,需要将指定服务器上的指定路径下的所有文件,例如:test.csv 上传到hdfs上,参考了网上的一些大佬的帖子实现了需求,故做个笔记方便以后查看,参考地址:https://www.cnblogs.com/SmallCaff/p/10650699.html
import os
import platform
import logging as lg
from hdfs.client import Client
from hdfs.client import _logger
'''
使用前需先安装
pip install hdfs
'''
#setting配置部分
HOST = "http://192.168.12.101:50070" # 地址
ROOT_PATH = "/" # 客户端连接的目录
REMOTE_PATH = "/air_test_data" # HDFS上的路径,注意,需要先在hdfs手动创建此目录
LOCAL_PATH = "file/excel_file" # 本地路径,建议写绝对路径,例如:E:\my_work\测试目录
LOG_DIR = "log/upload.log" # 日志路径 建议写绝对路径,例如:E:\my_work\log
#-------------------------------------------------------
# 记录日志(用了hdfs中源码的logger)
logdir = LOG_DIR
_logger.setLevel(lg.DEBUG)
myhandler = lg.FileHandler(logdir, "w") # 覆盖模式输出日志
myhandler.setLevel(lg.DEBUG)
formatter = lg.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
myhandler.setFormatter(formatter)
_logger.addHandler(myhandler)
class Transfer(object):
def __init__(self):
self.host = HOST
self.remotepath = REMOTE_PATH
self.localpath = LOCAL_PATH
self.rootpath = ROOT_PATH
self.client = Client(self.host, root=self.rootpath)
def upload_file_windows(self):
"""windowsserver"""
try:
base_dir = self.localpath.split('\\').pop() # 要上传的路径的最后一个文件夹
for root, dirs, files in os.walk(self.localpath):
new_dir = base_dir + root.split(base_dir).pop().replace('\\','/') # 去除本地路径前缀
for file in files:
old_path = root + '\\' + file # 原始本地路径文件
lpath = new_dir + '/' + file # 去除本地路径前缀后的文件
if not self.client.status(self.remotepath + '/' + lpath, strict=False):
# 第一个参数远程路径,第二个参数本地路径,第三个参数是否覆盖,第四个参数工作线程数
self.client.upload(self.remotepath + '/' + lpath, old_path, overwrite=False)
except Exception as e:
with open("err.log", "a") as f:
f.write(str(e))
def upload_file_linux(self, sep):
"""linuxserver"""
try:
base_dir = self.localpath.split(sep).pop() # 要上传的路径的最后一个文件夹
for root, dirs, files in os.walk(self.localpath):
new_dir = base_dir + root.split(base_dir).pop() # 去除本地路径前缀
for file in files:
old_path = root + sep + file # 原始本地路径文件
lpath = new_dir + sep + file # 去除本地路径前缀后的文件
if not self.client.status(self.remotepath + sep + lpath, strict=False):
# len(old_path.split("/"))
# 第一个参数远程路径,第二个参数本地路径,第三个参数是否覆盖,第四个参数工作线程数
self.client.upload(self.remotepath + sep + lpath, old_path, overwrite=False)
except Exception as e:
with open("err.log", "a") as f:
f.write(str(e))
if __name__ == "__main__":
transfer = Transfer()
# windows server
if platform.platform().startswith("Windows"):
transfer.upload_file_windows()
else:
# linux server
transfer.upload_file_linux('/')
上传效果: