def run_hive(configData: ConfigData):
a_client = InsecureClient(url=configData.hdfs_ip(), user="******") # "http://10.2.201.197:50070"
conn = connect(host=configData.hive_ip(), port=configData.hive_port(), auth_mechanism=configData.hive_auth(), user=configData.hive_user())
cur = conn.cursor()
f_date_str = configData.get_f_date() # "20181101"
p_date_str = configData.get_p_date() # "2018-11-01"
# hdfs_dir_bl
root_path = str(pathlib.PurePosixPath(configData.get_hdfs_path()).joinpath(f_date_str))
file_name = str(pathlib.PurePosixPath(root_path).joinpath(configData.get_file_name(f_date_str)))
# "/data/posflow/allinpay_utf8_zc/20181101/"
# 20181101_loginfo_rsp_bl_new.csv
# 20181101_rsp_agt_bl_new.del
# 20181101_rxinfo_rsp_bl.txt
table_name = configData.get_table_name()
print("Start\n")
if MyHdfsFile.isfile(a_client, file_name):
if not configData.get_has_partition():
sql = "LOAD DATA INPATH '{}' INTO TABLE {}".format(file_name, table_name) # 'test.t1_trxrecprd_v2_zc'
# '\' OVERWRITE INTO TABLE test.t1_trxrecprd_v2_bl2'
else:
sql = "LOAD DATA INPATH '{}' INTO TABLE {} PARTITION ( p_date='{}' )".format(file_name, table_name, p_date_str) # 'test.t1_trxrecprd_v2_zc'
print("OK" + " " + sql+"\n")
cur.execute(sql) # , async=True)
cur.close()
conn.close()
def run_hive(configData: ConfigData):
a_client = InsecureClient(url=configData.hdfs_ip(), user="******") # "http://10.2.201.197:50070"
conn = connect(host=configData.hive_ip(), port=configData.hive_port(), auth_mechanism=configData.hive_auth(), user=configData.hive_user())
cur = conn.cursor()
f_date_str = configData.get_f_date() # "20181101"
p_date_str = configData.get_p_date() # "2018-11-01"
root_path = configData.get_hdfs_path() # "/shouyinbao/bl_shouyinbao/UTF8/"
file_name = configData.get_file_name(f_date_str) # "t1_trxrecord_" the_date # "_V2.csv"
table_name = configData.get_table_name()
print("Start\n")
idn = 0
branches = MyHdfsFile.get_child(a_client, str(pathlib.PurePosixPath(root_path).joinpath(f_date_str)))
for aBranch in branches:
if MyHdfsFile.check_branch(a_client, aBranch):
files = MyHdfsFile.get_child(a_client, aBranch)
f_a_branch = MyHdfsFile.get_name(aBranch)
for aFile in files:
if MyHdfsFile.check_file(a_client, aFile, file_name):
# '/shouyinbao/bl_shouyinbao/UTF8/20181101/9999997900/t1_trxrecord_20181101_V2.csv'
to_file2 = str(pathlib.PurePosixPath(root_path).joinpath(f_date_str, f_a_branch, file_name))
if not configData.get_has_partition():
sql = "LOAD DATA INPATH '{}' INTO TABLE {}".format(to_file2, table_name) # 'test.t1_trxrecprd_v2_zc'
# '\' OVERWRITE INTO TABLE test.t1_trxrecprd_v2_bl2'
else:
sql = "LOAD DATA INPATH '{}' INTO TABLE {} PARTITION ( p_date='{}' )".format(to_file2, table_name, p_date_str) # 'test.t1_trxrecprd_v2_zc'
idn += 1
print(str(idn) + " " + sql + "\n")
cur.execute(sql) # , async=True)
cur.close()
conn.close()
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)
def run_remove_files(configData: ConfigData):
f_date_str = configData.get_f_date() # StrTool.get_the_date_str(the_date, delta_day) # "20181101"
data_path = os.path.join(configData.get_data_path(), f_date_str) # allinpay_data_bl
utf8_path = os.path.join(configData.get_utf8_path(), f_date_str) # allinpay_utf8_bl
hdfs_path = str(pathlib.PurePosixPath(configData.get_hdfs_path()).joinpath(f_date_str)) # hdfs_dir_bl
a_client = InsecureClient(configData.hdfs_ip(), user="******") # "http://10.2.201.197:50070"
shutil.rmtree(data_path, ignore_errors=True)
shutil.rmtree(utf8_path, ignore_errors=True)
try:
a_client.delete(hdfs_path, recursive=True)
except:
pass
def run_conv_file_local_to_hdfs(configData: ConfigData):
"""
:param configData:
:return:
"""
f_date_str = configData.get_f_date() # "20181101"
a_client = InsecureClient(configData.hdfs_ip(), user="******") # "http://10.2.201.197:50070"
root_path = os.path.join(configData.get_data_path(), f_date_str) # allinpay_data_bl
dest_dir1 = os.path.join(configData.get_utf8_path(), f_date_str) # allinpay_utf8_bl
dest_dir2 = str(pathlib.PurePosixPath(configData.get_hdfs_path()).joinpath(f_date_str)) # hdfs_dir_bl
# file_ext7 = configData.get_data("file_ext7") # _loginfo_rsp_bl_new.csv # 20181101_loginfo_rsp_bl_new.csv
# file_ext8 = configData.get_data("file_ext8") # _rsp_agt_bl_new.del # 20181101_rsp_agt_bl_new.del
# file_ext9 = configData.get_data("file_ext9") # _rxinfo_rsp_bl.txt # 20181101_rxinfo_rsp_bl.txt
# f_list = [file_ext7, file_ext8, file_ext9]
print("Start\n")
# "file_ext" + str(configData.the_id)
file_name = configData.get_file_name(f_date_str).lower()
files = MyLocalFile.get_child_file(root_path)
for aFile in files:
short_name = os.path.basename(aFile).lower()
f_name = pathlib.PurePath(aFile).name
if short_name == file_name:
to_file1 = str(pathlib.PurePath(dest_dir1).joinpath(f_name))
to_file2 = str(pathlib.PurePosixPath(dest_dir2).joinpath(f_name))
f_add_date = configData.get_hive_add_date(f_date_str)
f_need_head = not configData.get_hive_head()
MyLocalFile.conv_file_local(aFile, to_file1, need_first_line=f_need_head, p_add_head=f_add_date)
MyHdfsFile.safe_make_dir(a_client, to_file2)
# a_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)
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)
def run_conv_file_hdfs(configData: ConfigData):
f_date_str = configData.get_f_date() # "20181101"
client = InsecureClient(configData.hdfs_ip(), user="******") # "http://10.2.201.197:50070"
root_path = configData.get_data_path() # 'D:/DATA/UNZIP/'
dest_dir = configData.get_hdfs_path()
f_name = configData.get_file_name(f_date_str) # "t1_trxrecord_" the_date # "_V2.csv"
print("Start\n")
branches = MyLocalFile.get_child(os.path.join(root_path, f_date_str))
for aBranch in branches:
if MyLocalFile.check_branch(aBranch):
files = MyLocalFile.get_child(aBranch)
for aFile in files:
if MyLocalFile.check_file(aFile, f_name):
MyHdfsFile.conv_file_hdfs(aFile,
os.path.join(dest_dir,
f_date_str,
os.path.basename(aBranch),
f_name),
client)