def client_init(user):
session = requests.session()
session.keep_alive = False
client = pyhdfs.HdfsClient(hosts="192.168.7.150:9870",
user_name=user,
requests_session=session)
return client
def __init__(self):
hosts = ",".join(
["{}:{}".format(host, HDFS_PORT) for host in HDFS_HOST])
self.client = pyhdfs.HdfsClient(hosts=hosts,
user_name="szliu",
max_tries=5,
retry_delay=5)
def CCP():
fs = pyhdfs.HdfsClient(hosts="127.0.0.1:50070", user_name="kali")
f = fs.open("/user/kali/lab2/outputCCP/part-00000")
arr = []
for line in f:
strin = ", ".join(line.decode()[0:-1].split("\t"))
pairs = strin.split(", ")
pairs[2] = int(pairs[2])
arr.append(pairs)
arr.sort(key=lambda x: x[2], reverse=True)
# for item in arr:
# print(item)
check_word = input("Write item's name: ")
count = 0
for i in range(len(arr)):
if check_word.lower() == arr[i][0].lower():
count += 1
print(f"{count}:\t{arr[i][1]}: {arr[i][2]}")
elif check_word.lower() == arr[i][1].lower():
count += 1
print(f"{count}:\t{arr[i][0]}: {arr[i][2]}")
if count == 10:
break
def SchemaExtractor(hdfs_host=HDFS_HOST, hdfs_port=HDFS_IPC_PORT, path=None):
''' Infer the schema of a Parquet file from its metadata
:param hdfs_host:
:param hdfs_port:
:param path: path to the parquet file
:return: schema (String)
'''
sqlContext = SQLContext(sc)
seperator = ':'
hdfs_path = [hdfs_host, str(hdfs_port)]
fs = pyhdfs.HdfsClient(hosts=seperator.join(hdfs_path),
user_name='spark24')
file_list = []
for root, _, filenames in fs.walk(path):
for filename in filenames:
if filename.endswith(".parquet"):
file_list.append(root + '/' + filename)
single_parquet_path = file_list[0]
df = sqlContext.read.parquet('hdfs://' + hdfs_host + single_parquet_path)
schema = ''
for name, dtype in df.dtypes:
schema = schema + ' ' + str(name) + ' ' + str(dtype) + ','
schema = schema[:-1]
return schema
def handleFileFromHdfs(fileName,
rootFolder,
jsonData={},
userName='******',
hdfsHost="spark-master0",
nnPort="50070"):
'''
remove or rename the file from hdfs
'''
try:
client = pyhdfs.HdfsClient(hosts="{0}:{1}".format(hdfsHost, nnPort))
except Exception:
logger.error("Exception: {0}, Traceback: {1}".format(
sys.exc_info(), traceback.format_exc()))
return False
if not jsonData:
'''
check the generated file if exist
'''
FolderUri = "{0}/{1}/{2}".format(rootFolder, userName, fileName)
if client.exists(FolderUri):
return True
return False
if rootFolder.startswith('/tmp/users'):
csvFolderUri = "{0}/{1}/csv/{2}".format(rootFolder, userName, fileName)
parquetFolderUri = "{0}/{1}/parquet/{2}".format(
rootFolder, userName, fileName)
if jsonData['method'] == 'delete':
deleteHdfsFile(client, csvFolderUri)
deleteHdfsFile(client, parquetFolderUri)
return True
elif jsonData['method'] == 'rename':
newname = jsonData['newname']
csvRs = renameHdfsFile(client, csvFolderUri, newname)
parquetRs = renameHdfsFile(client, parquetFolderUri, newname)
if csvRs and parquetRs:
return True
return False
elif rootFolder.startswith('/users'):
username = jsonData['username'] if 'username' in jsonData.keys(
) else 'yzy'
FolderUri = "{0}/{1}/{2}".format(rootFolder, userName, fileName)
if jsonData['method'] == 'delete':
checkOrDeleteView(fileName, username, delete=True)
return deleteHdfsFile(client, FolderUri)
elif jsonData['method'] == 'rename':
newname = jsonData['newname']
return renameHdfsFile(client,
FolderUri,
newname,
username=username)
def seplunk_start(config_path):
if not os.path.exists(config_path):
LOGGER.error("config file path not exists")
sys.exit(1)
config = get_configure(config_path)
default_fs = config.get('fs.defaultFS', None)
if not default_fs:
LOGGER.error("hdfs not found")
host = default_fs.split(':')[1].strip("/")
hdfs_client = pyhdfs.HdfsClient(host)
if not os.path.exists(TMP_PATH):
os.mkdir(TMP_PATH)
db_path = os.path.join(TMP_PATH, DB_FLIE)
# if os.path.exists(db_path):
# os.remove(db_path)
conn = sql.get_conn(db_path)
p_log_conn, p_monitor_conn = Pipe()
p_log = Process(target=create_process_log,
args=(hdfs_client, config, conn, p_log_conn))
p_log.start()
p_monitor = Process(target=create_process_monitor,
args=(hdfs_client, config, conn, p_monitor_conn))
p_monitor.start()
def hdfs_get_file(remote_path,
filename,
local_path,
delete=False,
hdfs_path='10.20.37.175',
port=9000):
hdfs_client = pyhdfs.HdfsClient(hdfs_path, port)
try:
hdfs_client.copy_to_local(remote_path + filename,
local_path + filename)
except:
time.sleep(2)
hdfs_client.copy_to_local(remote_path + filename,
local_path + filename)
print("load completed!")
if delete:
files = hdfs_client.listdir(remote_path)
while filename in files:
try:
hdfs_client.delete(remote_path + filename)
files = hdfs_client.listdir(remote_path)
time.sleep(2)
except:
files = hdfs_client.listdir(remote_path)
return local_path + filename
def _init_client(self):
s = requests.session()
s.keep_alive = False
client = pyhdfs.HdfsClient(hosts=HDFS_HOSTS,
user_name=self.user,
requests_session=s)
return client
def up_hdfs(self,new_file):
r"""
上传到hadoop上
"""
#链接hdfs
client = pyhdfs.HdfsClient(hosts='hadoop2x-01:50070,hadoop2x-02:50070',user_name='suh')
now_time = datetime.datetime.now().strftime("%Y%m%d")
year = time.strftime('%Y',time.localtime(time.time()))
HdfsDir = r'/RawData/aiaa_meeting/big_json/%s/%s'%(year,now_time)
if not client.exists(HdfsDir):
client.mkdirs(HdfsDir)
print('Before !%s' % client.listdir(HdfsDir))
#从本地上传文件至集群
local_path = new_file
up_path = HdfsDir + '/%s.big_json' % (now_time)
big_json_size = os.path.getsize(local_path)
if big_json_size != 0 :
if client.exists(up_path):
client.delete(up_path)
client.copy_from_local(local_path,up_path)
print('After !%s' % client.listdir(HdfsDir))
msg = '成功上传到%s' % HdfsDir
self.msg2weixin(msg)
else:
print("无更新，不上传")
def get_a_file():
fs = pyhdfs.HdfsClient(hosts='localhost:9870', user_name='hadoop')
lines = list()
with fs.open('/output/part-00000') as f:
for line in f:
lines.append(line.decode("utf-8").strip('\n'))
return lines
def main(keywords: str, tag_type: str) -> None:
"""
main function to get data from postgres and move it into hdfs
Parameters
----------
keywords : str
tag string
tag_type : str
partition in hive
See Also
----------
get_data : get poi related data from POI table
"""
res = get_data(keywords)
with TemporaryDirectory() as dirname:
# generate txt file
res.to_csv(os.path.join(dirname, f'POI码表_{keywords}_每日扫描版.txt'),
header=None,
index=None,
sep=',',
mode='w')
# move txt file into hdfs
hdfs_path = f'/user/hive/warehouse/poi.db/code_aoi_geohash/tag_type={tag_type}/POI码表_{keywords}_每日扫描版.txt'
hdfs = pyhdfs.HdfsClient(['10.244.16.101', '10.244.16.102'],
user_name='hdfs')
hdfs.copy_from_local(os.path.join(dirname,
f'POI码表_{keywords}_每日扫描版.txt'),
hdfs_path,
overwrite=True)
def file_upload(self, host, user_name, local_path, hdfs_path):
print("===== file upload start =====")
fs = pyhdfs.HdfsClient(hosts=host, user_name=user_name)
print(fs.get_active_namenode())
print(fs.listdir('/'))
fs.copy_from_local(local_path, hdfs_path)
print("==== file upload finish =====")
def upload_file_to_hdfs(hdfs_path, src_path):
"""
上传文件到HDFS文件系统，如果文件夹不存在就创建
Arguments:
hdfs_path {string} -- HDFS文件夹
src_path {string} -- 源地址文件夹
Raises:
ValueError -- 当`hdfs_path`不是类似`Linux`文件路径就会抛出
e -- `pyhdfs`模块的异常
"""
if not hdfs_path.startswith("/"):
raise ValueError(
'属性`hdfs_path`必须为有效的hadoop平台路径，例如 /RawData/chaoxing/duxiu_ts/big_htm'
)
name_node = cf.get('hadoop', 'namenode')
try:
client = pyhdfs.HdfsClient(hosts=name_node)
if not client.exists(hdfs_path): # 如果hadoop平台文件夹不存在，就创建文件夹
client.mkdirs(hdfs_path)
for _, local_file in file_list(src_path):
filename = os.path.basename(local_file)
dst_file = hdfs_path + '/' + filename
client.copy_from_local(local_file, dst_file, overwrite=True)
except pyhdfs.HdfsException as e:
raise e
def __init__(self, spark):
self.spark = spark
self.precision_span = 60
self.timestack_span = 600
self.HDFS_CLIENT = pyhdfs.HdfsClient(hosts=hdfs_host+':9870', user_name='hdfs')
self.HDFS_AUDIT = hdfs_audit_path
self.HDFS_IM = hdfs_im_path
def client_init(hosts):
session = requests.session()
session.keep_alive = False
client = pyhdfs.HdfsClient(hosts=hosts,
user_name="hdfs",
requests_session=session)
return client
def __init__(self, host: str = "localhost", port: int = 9870, user_name="root", **kwargs):
super().__init__()
self.host = host
self.port = port
self.user = user_name
self.home = f"/user/{self.user:s}"
self.fs = pyhdfs.HdfsClient(
hosts=[f"{host:s}:{port:d}", "localhost:9870"], user_name=user_name, **kwargs)
def get_conn(hosts,user): """ get a client of hadoop distributed file system parameter : 'hosts' is like 'host1:port1,host2:port2', use ',' to divide different server 'user' is the user of hadoop """ return pyhdfs.HdfsClient(hosts=hosts,user_name=user)
def up_hdfs(self):
#链接hdfs
client = pyhdfs.HdfsClient(hosts='hadoop2x-01:50070,hadoop2x-02:50070',user_name='suh')
# 从本地上传文件至集群之前，集群的目录
print('Before !%s' % client.listdir('/RawData/asceproceedings/big_json/2019/20190812'))
#从本地上传文件至集群
client.copy_from_local(r'D:\code\proceedings\big_json\20190812_1.big_json','/RawData/asceproceedings/big_json/2019/20190812/20190812_1.big_json')
# 从本地上传文件至集群之后，集群的目录
print('After !%s' % client.listdir('/RawData/asceproceedings/big_json/2019/20190812'))
def savetohdfs(rdd):
client = pyhdfs.HdfsClient(hosts="10.120.14.120,50070",
user_name="cloudera")
for r in rdd:
for t in r:
client.append(
"/user/cloudera/model_deploy/output/utime.csv",
"This patient need {} seconds to detect from MRI\n".format(
str(t)))
def __init__(self, spark):
self.spark = spark
self.HDFS_CLIENT = pyhdfs.HdfsClient(hosts=HDFS_HOST + ':9870',
user_name='hdfs')
self.HDFS_AUDIT = HDFS_AUDIT_COMPRESS_PATH
self.HDFS_IM = HDFS_IM_COMPRESS_PATH
self.probe_fields_change = [['start_time', 'probe_time']]
self.wifi_fields_change = [['collect_time', 'probe_time'],
['wifi_mac', 'probe_mac']]
def hdfs_init_total(remote_path, hdfs_path="10.20.37.175", port=9000):
hdfs_client = pyhdfs.HdfsClient(hdfs_path, port)
try:
files = hdfs_client.listdir(remote_path)
for k in files:
hdfs_client.delete(remote_path + k)
hdfs_client.delete(remote_path)
except:
return
def judgeIcon(request,
hdfsHost="spark-master0",
nnPort="50070",
csvUrl="/tmp/users",
url="/users",
userName="******"):
'''
返回图标状态（是否亮）
'''
jsonData = request.data
logger.debug('jsondata: {0}'.format(jsonData))
if request.method == 'POST':
username = jsonData['username'] if 'username' in jsonData.keys(
) else 'yzy'
second = 0
third = 0
fourth = 1
if username in Singleton().dataPaltForm.keys():
for key, value in Singleton().dataPaltForm[username].items():
if judgeConn(value.con):
second = 1
break
csvUrl = '{0}/{1}'.format(csvUrl, userName)
url = '{0}/{1}'.format(url, userName)
client = pyhdfs.HdfsClient(hosts="{0}:{1}".format(hdfsHost, nnPort))
if client.exists(url):
fileList = client.listdir(url)
if fileList:
second = 1
third = 1
if second == 0:
if client.exists('{0}/{1}'.format(csvUrl, 'parquet')):
parquetList = client.listdir('{0}/{1}'.format(
csvUrl, 'parquet'))
if parquetList:
second = 1
folderList = DashboardFolderByUser.objects.filter(username=username)
viewList = DashboardViewByUser.objects.filter(username=username)
indexList = DashboardIndexByUser.objects.filter(username=username)
logger.debug('folderList: {0}, viewList: {1}, indexList: {2}'.format(
folderList, viewList, indexList))
if (len(folderList) == 0 and len(viewList) == 0
and len(indexList) == 0) or third == 0:
fourth = 0
context = {
"status": "success",
"results": {
"constructview": second,
"dashboardview": third,
"statementview": fourth
}
}
return JsonResponse(context)
def update_track_talbe(self, df_track_output):
"""执行更新操作，将需要更新的数据列出，将新的数据追加到track表中"""
# 旧track表的数据
# df_track_old[id,probe_time,probe_type,probe_data,probe_device_mac,netbar_wacode,longitude,latitude,device_type,create_time,datasource_id,datasource_table_name,base_person_id,flag]
# 到时候df_track_old读取的数据（track表）维护在hdfs里，最后写回hdfs与pgsql
client = pyhdfs.HdfsClient(hosts=self.PYHDFS_HOST, user_name='hdfs')
if client.exists(self.HDFS_ST_TRACK_INFO_SUB):
df_track_old = self.spark.read.parquet(self.HDFS_ST_TRACK_INFO)
else:
df_track_old = self.spark.read.jdbc(
url=self.PGSQL_URL_analysis,
table=self.PGSQL_TABLE_TRACK,
properties=self.PGSQL_PROPERTIES).limit(0).withColumn(
"flag", functions.lit(0))
# df_track_old = self.spark.read.csv(self.HDFS_ST_TRACK, head=True)
df_track_old = df_track_old[df_track_old["datasource_table_name"] ==
"gd_ele_fence"]
df_track_old = df_track_old.withColumn(
"device_type", df_track_old.device_type.astype("long"))
# _df_track_old [netbar_wacode,probe_time,probe_data,probe_type,datasource_table_name,longitude,latitude,device_type,probe_device_mac,flag]
_df_track_old = df_track_old.select("netbar_wacode", "probe_time",
"probe_data", "probe_type",
"datasource_table_name",
"longitude", "latitude",
"device_type", "probe_device_mac",
"flag") # , "last_update_time"
# df_track_output[netbar_wacode,probe_time,probe_data,probe_type,create_time,datasource_table_name,longitude,latitude,device_type,probe_device_mac,flag]
df_track_output = df_track_output.drop("create_time").drop("count")
# # df_track_output[netbar_wacode,probe_time,probe_data,probe_type,datasource_table_name,longitude,latitude,device_type,probe_device_mac,flag]
# # 旧表中有的去掉新表中有的，留下旧表中已被更新的数据，以供update
# # 由于需要update的数据在append中也有，所以需要将append中的update的数据进行剔除
# df_track_update = _df_track_old.subtract(df_track_output)
# # df_track_update[netbar_wacode,probe_time,probe_data,probe_type,datasource_table_name,longitude,latitude,device_type,probe_device_mac,flag,create_time]
# df_track_update = df_track_update.withColumn("create_time", functions.lit(int(time.time() * 1000)))
# df_track_update = df_track_update.join(df_track_old.select("id", "probe_time", "probe_data", "probe_device_mac", "flag"), ["probe_time", "probe_data", "probe_device_mac", "flag"], "left")
# df_track_update.select("id", "probe_time").show()
# # 到时候和旧track进行join，取到id，之后通过id进行更新time_on，即time_end
# subtract 会自动去重掉原df的数据
# ！！！怎么解决？ df中维护start和end
# 当结构两个df结构没有完全相同，例如nullable，subtract不会求差
df_track_append = df_track_output.subtract(_df_track_old)
# df_track_output.show()
# _df_track_old.show()
df_track_append = df_track_append.withColumn(
"device_type", df_track_append.device_type.astype("long"))
df_track_append = df_track_append.withColumn(
"create_time", functions.lit(int(time.time() * 1000)))
# 此为有flag版，到时候插入轨迹表需要将flag字段去掉
# df_track_append.write.mode(saveMode='append').jdbc(url=self.PGSQL_URL_analysis, table="zhaoqing_duanzhou_db.track_copy2", properties=self.PGSQL_PROPERTIES)
df_track_append.drop("flag").write.mode(saveMode='append').jdbc(
url=self.PGSQL_URL_analysis,
table=self.PGSQL_TABLE_TRACK,
properties=self.PGSQL_PROPERTIES)
df_track_append.write.parquet(self.HDFS_ST_TRACK_INFO, mode="append")
def __init__(self):
conf = SparkConf().setAppName('deviceupdate')
conf.set("spark.executor.memory", "3g")
conf.set("spark.driver.memory", "3g")
conf.set("spark.executor.core", "3")
self.spark = SparkSession.builder.config(conf=conf).getOrCreate()
self.PGSQL_URL = f"jdbc:postgresql://{DST_DB_HOST}:{DST_DB_PORT}/{DST_DB_DATABASE}"
self.PGSQL_DEVICE_TABLE = DST_DB_SCHEMA+".gd_device"
self.PGSQL_PLACE_TABLE = DST_DB_SCHEMA+".gd_place"
self.PGSQL_PROPERTIES = {'user': DST_DB_USER, 'password': DST_DB_PASSWORD}
self.HDFS_CLIENT = pyhdfs.HdfsClient(hosts=HDFS_HOST+':9870', user_name='hdfs')
def hdfs_test():
fs = pyhdfs.HdfsClient("localhost", 9000)
f = fs.open('/common_step3/part-00000')
for i in f:
line = str(i, encoding='utf-8')
# print(j)
rules, confidence = line.split('\t')
left_rules, right_rules = rules.split('->')
print('{0}---->{1}---->{2}'.format(left_rules, right_rules,
float(confidence.replace('\n',
''))))
def load_wordlist(filename):
hdfs = pyhdfs.HdfsClient(hosts='hdfs-namenode:50070')
words = {}
f = hdfs.open(filename)
text = f.read().decode('utf-8')
text = text.split('\n')
for line in text:
words[line] = 1
f.close()
return words
def hdfs_set_file(local_file_path,
remote_file_path,
filename,
hdfs_path="10.20.37.175",
port=9000):
hdfs_client = pyhdfs.HdfsClient(hdfs_path, port)
files = hdfs_client.listdir(remote_file_path)
if filename in files:
hdfs_client.delete(remote_file_path + filename)
hdfs_client.copy_from_local(local_file_path + filename,
remote_file_path + filename)
print("set Completed!")
def hdfs_get_file(remote_path,
filename,
local_path,
delete=False,
hdfs_path='10.20.37.175',
port=9000):
hdfs_client = pyhdfs.HdfsClient(hdfs_path, port)
hdfs_client.copy_to_local(remote_path + filename, local_path + filename)
print("load completed!")
if delete:
hdfs_client.delete(remote_path + filename)
return local_path + filename
def _save(self, name, content):
'''保存文件时用'''
# name，选择上传文件的名字
# content,包含上传文件内容的File对象
# 创建一个hdfs_client对象
client = pyhdfs.HdfsClient(hosts=self.hosts, user_name=self.user_name)
#判断文件是否上传，如果没有上传就上传
if not client.exists(self.file_url + content.name):
client.create(self.file_url + content.name, content)
#否则文件已上传则不用上传，直接返回文件名
filename = self.file_url + content.name
return filename
def hdfs_init_fold(remote_path, hdfs_path="10.20.37.175", port=9000):
hdfs_client = pyhdfs.HdfsClient(hdfs_path, port)
try:
files = hdfs_client.listdir(remote_path)
except:
hdfs_client.mkdirs(remote_path)
return
if files == []:
return
else:
for k in files:
hdfs_client.delete(remote_path + k)
return
