def run_remove_hive(conf: ConfigData, the_date: str, delta_day=0):
f_date_str = StrTool.get_the_date_str(the_date, delta_day) # "20181101"
# "/user/hive/warehouse/rds_posflow.db/t1_trxrecprd_v2/t1_trxrecord_20181204_V2*.csv"
del_table = conf.get_table_name(
) # hive_table="rds_posflow.t1_trxrecprd_v2"
if the_conf.m_project_id == 1:
del_file = conf.get_file_name(f_date_str).replace('.', '*.')
MyHdfsFile.delete_hive_ssh(conf.get_data("cdh_ip"),
table=del_table,
p_name=del_file,
username=conf.get_data("cdh_user"),
password=conf.get_data("cdh_pass"))
if the_conf.m_project_id == 2:
conn = connect(host=conf.hive_ip(),
port=conf.hive_port(),
auth_mechanism=conf.hive_auth(),
user=conf.hive_user())
cur = conn.cursor()
# "ALTER TABLE rds_posflow.t1_trxrecprd_v2_tmp DROP IF EXISTS PARTITION(p_date=20190208) "
sql = "ALTER TABLE {} DROP IF EXISTS PARTITION( p_date={} )".format(
del_table, the_date)
print(sql)
cur.execute(sql)
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_hive(conf: ConfigData, the_date: str, delta_day=0):
f_date_str = StrTool.get_the_date_str(the_date, delta_day) # "20181101"
del_table = conf.get_table_name(
) # "hive_table" + str(conf.the_id) # "rds_posflow.loginfo_rsp_bl"
del_file = conf.get_file_name(f_date_str).replace(
'.', '*.') # "file_ext" + str(conf.the_id)
MyHdfsFile.delete_hive_ssh(conf.get_data("cdh_ip"),
table=del_table,
p_name=del_file,
username=conf.get_data("cdh_user"),
password=conf.get_data("cdh_pass"))
def run_conv_file_local_to_hdfs(conf: ConfigData, the_date: str):
"""
:param conf:
:param the_date:
:return:
"""
the_date = StrTool.get_the_date_str(the_date)
a_client = MyClient(conf.hdfs_ip()) # "http://10.2.201.197:50070"
# allinpay_data_bl
data_path = os.path.join(conf.get_data_path(), the_date)
# allinpay_utf8_bl
dest_dir1 = os.path.join(conf.get_utf8_path(), the_date)
# hdfs_dir_bl
dest_dir2 = str(
pathlib.PurePosixPath(conf.get_hdfs_path()).joinpath(the_date))
# file_ext7 = conf.get_data("file_ext7") # _loginfo_rsp_bl_new.csv # 20181101_loginfo_rsp_bl_new.csv
# file_ext8 = conf.get_data("file_ext8") # _rsp_agt_bl_new.del # 20181101_rsp_agt_bl_new.del
# file_ext9 = conf.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(conf.the_id)
file_name = conf.get_file_name(the_date).lower()
files = MyLocalFile.get_child_file(data_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 = conf.get_hive_add_date(the_date)
f_need_head = conf.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_hive(conf: ConfigData, the_date: str):
client = Client(conf.hdfs_ip()) # "http://10.2.201.197:50070"
conn = connect(host=conf.hive_ip(), port=conf.hive_port(), auth_mechanism=conf.hive_auth(), user=conf.hive_user())
cur = conn.cursor()
the_date = StrTool.get_the_date_str(the_date) # "20181101"
root_path = conf.get_data("hdfs_dir_zc") # "/data/posflow/allinpay_utf8_zc/"
file_ext3 = conf.get_data("file_ext3") # _loginfo_rsp.txt # 20181101_loginfo_rsp.txt
file_ext4 = conf.get_data("file_ext4") # _loginfo_rsp_agt.txt # 20181101_loginfo_rsp_agt.txt
file_ext5 = conf.get_data("file_ext5") # _rxinfo_rsp.txt # 20181101_rxinfo_rsp.txt
file_ext6 = conf.get_data("file_ext6") # _rxinfo_rsp_agt.txt # 20181101_rxinfo_rsp_agt.txt
print("Start\n")
file3 = str(pathlib.PurePosixPath(root_path).joinpath(the_date + file_ext3))
file4 = str(pathlib.PurePosixPath(root_path).joinpath(the_date + file_ext4))
file5 = str(pathlib.PurePosixPath(root_path).joinpath(the_date + file_ext5))
file6 = str(pathlib.PurePosixPath(root_path).joinpath(the_date + file_ext6))
f_list = [file3,file4,file5,file6]
t_list = ["hive_table3", "hive_table4", "hive_table5", "hive_table6"]
for n in range(0,4):
if MyHdfsFile.isfile(client, f_list[n]):
sql = 'LOAD DATA INPATH \'' + f_list[n] + '\' INTO TABLE ' + conf.get_data(t_list[n]) # 'test.t1_trxrecprd_v2_zc'
# '\' OVERWRITE INTO TABLE test.t1_trxrecprd_v2_bl2'
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"
# 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(conf: ConfigData, the_date: str):
a_client = Client(conf.hdfs_ip()) # "http://10.2.201.197:50070"
conn = connect(host=conf.hive_ip(),
port=conf.hive_port(),
auth_mechanism=conf.hive_auth(),
user=conf.hive_user())
cur = conn.cursor()
print("Start\n")
the_date = StrTool.get_the_date_str(the_date) # "20181101"
# hdfs_dir_bl
root_path = str(
pathlib.PurePosixPath(conf.get_hdfs_path()).joinpath(the_date))
file_name = str(
pathlib.PurePosixPath(root_path).joinpath(
conf.get_file_name(the_date)))
# "/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 = conf.get_table_name()
if MyHdfsFile.isfile(a_client, file_name):
sql = 'LOAD DATA INPATH \'' + file_name + '\' INTO TABLE ' + table_name # 'test.t1_trxrecprd_v2_zc'
# '\' OVERWRITE INTO TABLE test.t1_trxrecprd_v2_bl2'
print("OK" + " " + sql + "\n")
cur.execute(sql) # , async=True)
cur.close()
conn.close()
def run_hive(conf: ConfigData, the_date: str, is_baoli=True):
p_client = Client(conf.hdfs_ip()) # "http://10.2.201.197:50070"
conn = connect(host=conf.hive_ip(),
port=conf.hive_port(),
auth_mechanism=conf.hive_auth(),
user=conf.hive_user())
cur = conn.cursor()
the_date = StrTool.get_the_date_str(the_date) # "20181101"
root_path = conf.get_hdfs_path() # "/shouyinbao/bl_shouyinbao/UTF8/"
f_name = conf.get_file_name(
the_date) # "t1_trxrecord_" the_date # "_V2.csv"
table_name = conf.get_table_name()
print("Start\n")
idn = 0
branches = MyHdfsFile.get_child(p_client, root_path + the_date)
for aBranch in branches:
if MyHdfsFile.check_branch(p_client, aBranch):
files = MyHdfsFile.get_child(p_client, aBranch)
f_a_branch = MyHdfsFile.get_name(aBranch)
for aFile in files:
if MyHdfsFile.check_file(p_client, aFile, f_name):
# '/shouyinbao/bl_shouyinbao/UTF8/20181101/9999997900/t1_trxrecord_20181101_V2.csv'
to_file2 = str(
pathlib.PurePosixPath(root_path).joinpath(
the_date, f_a_branch, f_name))
if conf.m_project_id == 1:
sql = 'LOAD DATA INPATH \'{}\' INTO TABLE {}'.format(
to_file2, table_name) # 'test.t1_trxrecprd_v2_zc'
# '\' OVERWRITE INTO TABLE test.t1_trxrecprd_v2_bl2'
elif conf.m_project_id == 2:
sql = 'LOAD DATA INPATH \'{}\' INTO TABLE {} PARTITION ( p_branch=\'{}\', p_date={} )'.format(
to_file2, table_name, f_a_branch,
the_date) # 'test.t1_trxrecprd_v2_zc'
idn += 1
print(str(idn) + " " + sql + "\n")
cur.execute(sql) # , async=True)
cur.close()
conn.close()
def run_remove_hive(configData: ConfigData):
f_date_str = configData.get_f_date() # "20181101"
p_date_str = configData.get_p_date() # "2018-11-01"
del_table = configData.get_table_name() # "hive_table" + str(configData.the_id) # "rds_posflow.loginfo_rsp_bl"
print(configData.cdh_ip()+del_table+f_date_str+configData.get_file_name(f_date_str)+configData.hive_ip())
if not configData.get_has_partition():
del_file = configData.get_file_name(f_date_str).replace('.', '*.') # "file_ext" + str(configData.the_id)
MyHdfsFile.delete_hive_ssh(configData.cdh_ip(), table=del_table, p_name=del_file, username=configData.cdh_user(), password=configData.cdh_pass())
else:
conn = connect(host=configData.hive_ip(), port=configData.hive_port(), auth_mechanism=configData.hive_auth(), user=configData.hive_user())
cur = conn.cursor()
# "ALTER TABLE rds_posflow.t1_trxrecprd_v2_tmp DROP IF EXISTS PARTITION(p_date='2019-02-08') "
sql = "ALTER TABLE {} DROP IF EXISTS PARTITION( p_date='{}' )".format(del_table, p_date_str)
print(sql)
cur.execute(sql)
cur.close()
conn.close()
def run_conv_file_hdfs(conf: ConfigData, the_date: str, is_baoli=True):
the_date = StrTool.get_the_date_str(the_date)
client = Client(conf.hdfs_ip()) # "http://10.2.201.197:50070"
root_path = conf.get_data_path() # 'D:/DATA/UNZIP/'
dest_dir = conf.get_hdfs_path()
f_name = conf.get_file_name(
the_date) # "t1_trxrecord_" the_date # "_V2.csv"
print("Start\n")
branches = MyLocalFile.get_child(os.path.join(root_path, the_date))
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, the_date,
os.path.basename(aBranch), f_name),
client)
def run_remove_hive(configData: ConfigData):
f_date_str = configData.get_f_date() # "20181101"
p_date_str = configData.get_p_date() # "2018-11-01"
# "/user/hive/warehouse/rds_posflow.db/t1_trxrecprd_v2/t1_trxrecord_20181204_V2*.csv"
del_table = configData.get_table_name() # hive_table="rds_posflow.t1_trxrecprd_v2"
if not configData.get_has_partition():
del_file = configData.get_file_name(f_date_str).replace('.', '*.')
MyHdfsFile.delete_hive_ssh(configData.cdh_ip(), table=del_table, p_name=del_file, username=configData.cdh_user(), password=configData.cdh_pass())
else:
conn = connect(host=configData.hive_ip(), port=configData.hive_port(), auth_mechanism=configData.hive_auth(), user=configData.hive_user())
cur = conn.cursor()
# "ALTER TABLE rds_posflow.t1_trxrecprd_v2_tmp DROP IF EXISTS PARTITION(p_date=2019-02-08) "
sql = "ALTER TABLE {} DROP IF EXISTS PARTITION( p_date='{}' )".format(del_table, p_date_str)
print(sql)
cur.execute(sql)
cur.close()
conn.close()
def run_conv_file_local_to_hdfs(conf: ConfigData, the_date: str):
"""
:param conf:
:param the_date:
:return:
"""
the_date = StrTool.get_the_date_str(the_date)
client = MyClient(conf.hdfs_ip()) # "http://10.2.201.197:50070"
root_path = conf.get_data("allinpay_data_zc")
dest_dir1 = conf.get_data("allinpay_utf8_zc")
dest_dir2 = conf.get_data("hdfs_dir_zc")
file_ext3 = conf.get_data("file_ext3") # _loginfo_rsp.txt # 20181101_loginfo_rsp.txt
file_ext4 = conf.get_data("file_ext4") # _loginfo_rsp_agt.txt # 20181101_loginfo_rsp_agt.txt
file_ext5 = conf.get_data("file_ext5") # _rxinfo_rsp.txt # 20181101_rxinfo_rsp.txt
file_ext6 = conf.get_data("file_ext6") # _rxinfo_rsp_agt.txt # 20181101_rxinfo_rsp_agt.txt
print("Start\n")
files = MyLocalFile.get_child_file(root_path)
for aFile in files:
short_name = os.path.basename(aFile).lower()
if short_name == (the_date + file_ext3).lower() or \
short_name == (the_date + file_ext4).lower() or \
short_name == (the_date + file_ext5).lower() or \
short_name == (the_date + file_ext6).lower():
to_file1 = str(pathlib.PurePath(dest_dir1).joinpath(pathlib.PurePath(aFile).name))
to_file2 = str(pathlib.PurePosixPath(dest_dir2).joinpath(pathlib.PurePath(aFile).name))
MyLocalFile.conv_file_local(aFile, to_file1, need_first_line=True)
MyHdfsFile.safe_make_dir(client, to_file2)
# client.newupload(to_file2, to_file1, encoding='utf-8')
the_file = client.status(to_file2, strict=False)
if the_file is None:
client.upload(to_file2, to_file1)
client.set_permission(to_file2, 777)
# client.set_owner(thePath,owner='hdfs',group='supergroup')
elif the_file['type'].lower() == 'file': # 'directory'
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)
hdfs_path = str(pathlib.PurePosixPath(conf.get_hdfs_path()).joinpath(sdate))
shutil.rmtree(data_path, ignore_errors=True)
shutil.rmtree(utf8_path, ignore_errors=True)
client = MyClient(conf.hdfs_ip()) # "http://10.2.201.197:50070"
client.delete(hdfs_path, recursive=True)
# "/user/hive/warehouse/posflow.db/t1_trxrecprd_v2/t1_trxrecord_20181204_V2*.csv"
# hive_table="posflow.t1_trxrecprd_v2",
# file_pre1 = 't1_trxrecord_',
# file_ext2 = "_V2.csv",
if __name__ == "__main__":
the_conf = ConfigData(p_is_test=False)
client = Client(the_conf.hdfs_ip()) # "http://10.2.201.197:50070"
a = MyHdfsFile.get_child(client, "/data/posflow/allinpay_utf8_zc")
b = MyHdfsFile.get_child_file(client,"/data/posflow/allinpay_utf8_zc")
c = MyHdfsFile.get_child_dir(client, "/data/posflow/allinpay_utf8_zc")
# test
# MyHdfsFile.delete(client, "/data/posflow/allinpay_utf8_zc", "*agt_cpy*")
# test
if the_conf.is_test():
day_str = the_conf.test_date()
days = 9
else:
day_str = StrTool.get_param_str(1, "")
days = StrTool.get_param_int(2, 1)
day_str = StrTool.get_the_date_str(day_str)