def basic():
client = HdfsClient(hosts='study:50070')
print(client.list_status('/'))
print '判断某个路径是否存在'
print client.exists("/test")
print client.exists("/data/gz/thrift-0.9.2.tar.gz")
client = HdfsClient(hosts='study:50070')
print client.get_file_checksum("/data/gz/bison-2.5.1.tar.gz")
summary = client.get_content_summary("/")
print summary
#文件拷贝--从HDFS拷贝到本地磁盘系统
client.copy_to_local("/data/gz/pip-7.1.2.tar.gz","/root/data/pip-7.1.2.tar.gz")
#文件拷贝--从本地磁盘系统拷贝到HDFS系统中
client.copy_from_local("/root/data/thrift-0.9.2.tar.gz","/data/gz/thrift-0.9.2.tar.gz")
print client.get_home_directory()
def start():
# 连接MongoDB，查询tokens，根据contractAddress到etherscan查询最新数据
client = MongoCluster().connect()
db = client.get_database('gse-transaction')
collection = db.get_collection('mrout_6000001-6001000')
# collection.insert_one()
# 连接HDFS读取文件
from pyhdfs import HdfsClient
client2 = HdfsClient(hosts='%s,50070' % hdfs_ip, max_tries=10)
# 返回这个用户的根目录
print client2.get_home_directory()
# 返回可用的namenode节点
print client2.get_active_namenode()
# 返回指定目录下的所有文件
print client2.listdir("/user/leon/mrout_3_6000001-6001000/")
# 读某个文件
client2.mkdirs("/user/leon")
inputfile = client2.open('/user/leon/mrout_3_6000001-6001000/part-00000')
# 查看文件内容
for r in inputfile:
line = str(r).encode('utf-8') # open后是二进制,str()转换为字符串并转码
print(line)
