def run_unzip_file(conf: ConfigData, the_date, folder_type=2):
the_date = StrTool.get_the_date_str(the_date)
if (type(the_date) is str) and len(the_date) == 8:
m_month = the_date[0:6]
m_day = the_date[6:8]
else:
return
zip_path = conf.get_zip_path()
data_path = conf.get_data_path()
f_name = conf.get_zip_name("") # "t1_trxrecord_" the_date # "_V2.csv"
print("Start\n")
# os.path.join(root_path, the_date) # real SYB folder don't have date folder
branches = MyLocalFile.get_child_dir(zip_path)
for aBranch in branches:
if MyLocalFile.check_branch(aBranch):
months = MyLocalFile.get_child_dir(aBranch)
for aMonth in months:
the_month = MyLocalFile.check_month(aMonth)
if the_month > 0 and "{:0>6d}".format(the_month) == m_month:
day_list = MyLocalFile.get_child_dir(aMonth)
for aDay in day_list:
the_day = MyLocalFile.check_day(aDay)
if the_day > 0 and "{:0>2d}".format(the_day) == m_day:
files = MyLocalFile.get_child_file(aDay)
for aFile in files:
if MyLocalFile.check_file(aFile,
p_name=f_name):
short_name = os.path.basename(aBranch)
if folder_type == 1:
new_path = os.path.join(
data_path, m_month, m_day,
short_name)
# "{:0>6d}".format(month) "{:0>2d}".format(day)
else:
new_path = os.path.join(
data_path, m_month + m_day,
short_name)
# "{:0>6d}{:0>2d}".format(month, day)
p_name = conf.get_file_name(m_month +
m_day)
MyLocalFile.unzip_the_file(
aFile, new_path, p_name)
def run_conv_file_local_to_hdfs(configData: ConfigData):
"""
# client.upload('/shouyinbao/', "/home/testFolder/logflow/bl_shouyinbao/UTF8/20181101/9999100000/t1_trxrecord_20181101_V2.csv", cleanup=True)
# dat = client.list('/shouyinbao/', status=False)
# print(dat)
# root_path = "/home/bd/桌面/201811_flow/zc_shouyinbao/UNZIP/"
# dest_dir1 = "/home/bd/桌面/201811_flow/zc_shouyinbao/UTF8/"
# dest_dir2 = "/shouyinbao/zc_shouyinbao/UTF8/"
# root_path = "/home/testFolder/logflow/bl_shouyinbao/UNZIP/"
# dest_dir1 = "/home/testFolder/logflow/bl_shouyinbao/UTF8/"
# dest_dir2 = "/shouyinbao/zc_shouyinbao/UTF8/"
# i_file = '/home/testFolder/logflow/bl_shouyinbao/20181101/9999100000/t1_trxrecord_20181101_V2.csv'
# o_file = '/home/testFolder/logflow/bl_shouyinbao/UTF8/20181101/9999100000/t1_trxrecord_20181101_V2.csv'
:param configData:
:param the_date:
:param is_baoli:
:return:
"""
f_date_str = configData.get_f_date() # "20181101"
a_client = InsecureClient(configData.hdfs_ip(), user="******") # "http://10.2.201.197:50070"
# webhdfs 默认是 dr.who ，不能伪装成其他用户，可以在配置里修改 hadoop.http.staticuser.user=dr.who
# https://www.cnblogs.com/peizhe123/p/5540845.html
root_path = os.path.join(configData.get_data_path(), f_date_str)
dest_dir1 = os.path.join(configData.get_utf8_path(), f_date_str)
dest_dir2 = str(pathlib.PurePosixPath(configData.get_hdfs_path()).joinpath(f_date_str))
print("Start\n")
f_name = configData.get_file_name(f_date_str) # "t1_trxrecord_" the_date # "_V2.csv"
branches = MyLocalFile.get_child_dir(root_path)
for aBranch in branches:
if MyLocalFile.check_branch(aBranch):
files = MyLocalFile.get_child_file(aBranch)
f_a_branch = os.path.basename(aBranch)
for aFile in files:
if MyLocalFile.check_file(aFile, f_name):
to_file1 = str(pathlib.PurePath(dest_dir1).joinpath(f_a_branch, f_name))
to_file2 = str(pathlib.PurePosixPath(dest_dir2).joinpath(f_a_branch, f_name))
f_add_head = configData.get_hive_add_date(f_a_branch)
f_add_end = configData.get_hive_add_date("789")
f_need_head = not configData.get_hive_head() # False
MyLocalFile.conv_file_local(aFile, to_file1, need_first_line=f_need_head,p_add_head=f_add_head, p_add_tail=f_add_end,quoting="")
MyHdfsFile.safe_make_dir(a_client, to_file2)
# client.newupload(to_file2, to_file1, encoding='utf-8')
the_file = a_client.status(to_file2, strict=False)
if the_file is None:
a_client.upload(to_file2, to_file1) #, encoding='utf-8')
a_client.set_permission(to_file2, 777)
# a_client.set_owner(thePath,owner='hdfs',group='supergroup')
elif the_file['type'].lower() == 'file': # 'directory'
a_client.set_permission(to_file2, 777)