def getReleaseFromHdfs_batch(releaseNumber=10):
'''copy releases form server using hdfs'''
client = hdfs.Client("http://10.141.221.82:50070", root='/home/fdse')
releases = client.list("/java/releases")
for r in releases[:releaseNumber]:
client.download(u'/java/releases/' + r, os.path.join(path, 'releases'))
print r, ' has download...'
def __init__(self,
uri='http://hadoop-hd1:50070',
user='******',
root='/user/hive/warehouse'):
self.client = hdfs.Client(uri)
self.user = user
self.root = root
def spark_df_to_local_txt(local_path,
df: pyspark.sql.DataFrame = None,
deli="\t",
hdfs_dir_path=None):
def row_to_str(row):
return deli.join([str(xx) for xx in row.asDict().values()])
# map as string rdd
if hdfs_dir_path == None:
str_rdd = df.rdd.map(row_to_str)
rand_bits = random.getrandbits(64)
hdfs_dir_name = "str_df_%016x" % rand_bits
hdfs_dir_path = "/tmp/" + hdfs_dir_name
# save to hdfs
str_rdd.saveAsTextFile(hdfs_dir_path)
else:
hdfs_dir_name = os.path.basename(hdfs_dir_path)
hdfs_dir_path = "/tmp/" + hdfs_dir_name
# get hdfs to local
local_tmp_dir = "./tmp/" + hdfs_dir_name
c = hdfs.Client("http://soldier1:50070")
c.download(hdfs_dir_path, local_tmp_dir)
# delete tmp hdfs
c.delete(hdfs_dir_path, recursive=True)
# cat to one file
with open(local_path, 'wb') as outfile:
for tmp_f in os.listdir(local_tmp_dir):
fn = os.path.join(local_tmp_dir, tmp_f)
with open(fn, 'rb') as readfile:
shutil.copyfileobj(readfile, outfile)
# remove local tmp_file
shutil.rmtree(local_tmp_dir)
def __init__(self,
uri='http://two-stream-master-prod-02:50070',
user='******',
root='hdfs://nameservicestream/user/hive/warehouse'):
self.client = hdfs.Client(uri)
self.user = user
self.root = root
def put_hdfs(shcontext, filename, hdfs_path):
try:
os.environ['NLS_LANG'] = 'AMERICAN_AMERICA.AL32UTF8' #设置环境变量，避免乱码
sh_path = '/root/spoolsh/' + filename + '.sh'
txt_local_path = '/root/spooldata/' + filename + '.txt'
f = open(sh_path, 'wb+')
f.write(shcontext)
f.close()
print '写入pool脚本'
oscmd1 = 'chmod +x ' + sh_path
(res_status1, res_output1) = commands.getstatusoutput(oscmd1)
oscmd2 = 'sh ' + sh_path
(res_status2, res_output2) = commands.getstatusoutput(oscmd2)
print 'spool数据到本地'
txt_hdfs_path = hdfs_path + filename + '.txt'
client = hdfs.Client("http://192.10.86.31:50070",
root="/",
timeout=100,
session=False)
client.delete(hdfs_path, recursive=True)
client.upload(txt_hdfs_path, txt_local_path)
print '上传数据到hdfs'
oscmd3 = 'rm -f ' + txt_local_path
print '删除本地文件'
res3 = os.system(oscmd3)
return 'success'
except Exception, e:
print e
return e
def read_data_sets(train_dir,
fake_data=False,
one_hot=False,
dtype=dtypes.float32,
reshape=True,
validation_size=5000,
seed=None):
if fake_data:
def fake():
return DataSet([], [],
fake_data=True,
one_hot=one_hot,
dtype=dtype,
seed=seed)
train = fake()
validation = fake()
test = fake()
return base.Datasets(train=train, validation=validation, test=test)
TRAIN_IMAGES = 'train-images-idx3-ubyte.gz'
TRAIN_LABELS = 'train-labels-idx1-ubyte.gz'
TEST_IMAGES = 't10k-images-idx3-ubyte.gz'
TEST_LABELS = 't10k-labels-idx1-ubyte.gz'
HOST = "192.168.205.185"
NAME_NODE_PORT = 50070
client = hdfs.Client('http://{}:{}'.format(HOST, NAME_NODE_PORT))
with client.read(train_dir + "/" + TRAIN_IMAGES) as f:
train_images = extract_images(f)
with client.read(train_dir + "/" + TRAIN_LABELS) as f:
train_labels = extract_labels(f, one_hot=one_hot)
with client.read(train_dir + "/" + TEST_IMAGES) as f:
test_images = extract_images(f)
with client.read(train_dir + "/" + TEST_LABELS) as f:
test_labels = extract_labels(f, one_hot=one_hot)
if not 0 <= validation_size <= len(train_images):
raise ValueError(
'Validation size should be between 0 and {}. Received: {}.'.format(
len(train_images), validation_size))
validation_images = train_images[:validation_size]
validation_labels = train_labels[:validation_size]
train_images = train_images[validation_size:]
train_labels = train_labels[validation_size:]
options = dict(dtype=dtype, reshape=reshape, seed=seed)
train = DataSet(train_images, train_labels, **options)
validation = DataSet(validation_images, validation_labels, **options)
test = DataSet(test_images, test_labels, **options)
return base.Datasets(train=train, validation=validation, test=test)
def client(self, host='', port='', timeout=None):
host = host or HDFS_CONFIG['host']
port = port or HDFS_CONFIG['webport']
if self._client is None:
try:
self._client = hdfs.Client('http://%s:%s' % (host, port), timeout=timeout, session=rs)
except Exception, e:
output('hdfs Exception ' + str(e), logType='hdfs')
raise
def get_storage_file_list(self):
from ..settings import FLINK_SAVEPOINT_PATH_BACKEND_ADDRESS
import hdfs
storage_file_backed = hdfs.Client(FLINK_SAVEPOINT_PATH_BACKEND_ADDRESS,
timeout=10)
storage_file_list = self.get_storage_file_list_status(
storage_file_backed, 'savepoints', self.get_savepoint_path())
# storage_file_list = self.get_storage_file_list_status(FLINK_SAVEPOINT_PATH_BACKEND, 'checkponits',
# self.get_checkponit_path())
storage_file_list.sort(key=lambda x: x.modification_time, reverse=True)
return storage_file_list
def table_to_hdfs(d_info):
d_info = p_insert_log(d_info)
localtempfile = str(d_info.get('localpath')) + str(
d_info.get('source_proc')) + '.tmp'
localfile = str(d_info.get('localpath')) + str(
d_info.get('source_proc')) + '.lz4'
if d_info.get('syn_strategy') == 1:
hdfspath = str(d_info.get('target_proc'))
else:
hdfspath = str(d_info.get('target_proc')) + str(
d_info.get('acctday')) + '/'
try:
d_info['source_dns'] = getdns(d_info.get('source_dbname').upper())
source_db = create_engine(d_info.get('source_dns'))
source_conn = source_db.connect()
try:
seldata = list(
source_conn.execute(
text(d_info.get('sql_select').format(**d_info))))
except Exception, etp:
logging.error(etp)
raise Exception(etp)
finally:
source_conn.close()
with open(localtempfile, 'wb+') as f:
for row_data in seldata:
f.write(''.join(row_data).encode('utf-8') + '\n')
logging.info('数据已写入本地，开始压缩数据')
oscmd = 'lz4 ' + localtempfile + ' ' + localfile
(status, output) = commands.getstatusoutput(oscmd)
if status == 0:
try:
d_info['target_dns'] = getdns(
d_info.get('target_dbname').upper())
client = hdfs.Client(d_info.get('target_dns'),
root="/",
timeout=100,
session=False)
if not client.status(hdfspath, strict=False):
client.makedirs(hdfspath, permission=777)
print localfile, hdfspath
client.upload(hdfspath, localfile)
d_info['finish_flag'] = 'finish'
d_info['retcode'] = 'success'
except Exception, etp:
logging.error(etp)
raise Exception(etp)
def download_file(hdfs_location, file_local):
"""
:param hdfs_location: hdfs路径+文件名
:param file_local: 本地路径+文件名
"""
# 可以用 try finally提高可读性
client = hdfs.Client('http://172.39.8.61:50070', root='/', timeout=10)
try:
with client.read(hdfs_location) as r:
with open(file_local, 'wb') as f:
f.write(r.read())
except hdfs.util.HdfsError:
client = hdfs.Client('http://172.39.8.62:50070', root='/', timeout=10)
with client.read(hdfs_location) as r:
with open(file_local, 'wb') as f:
f.write(r.read())
except IOError as msg:
with open("err.log", "a") as f:
f.write(str(msg))
file_local = None
return file_local
def __init__(self):
try:
self.zk = KazooClient(hosts=self.zooQuorum)
self.zk.start()
if self.zk.exists(self.path):
data = self.zk.get(self.path)
ip = None
if self.nameNodeA[0] in data[0]:
ip = 'http://' + self.nameNodeA[1] + ':50070'
elif self.nameNodeB[0] in data[0]:
ip = 'http://' + self.nameNodeB[1] + ':50070'
self.client = hdfs.Client(ip, root='/')
except Exception as e:
print('%s kazooClient __init__ ERROR! %s' %
(datetime.datetime.now(), traceback.format_exc()))
def _init_connect(self):
"""连接hdfs服务器"""
# 先获取hdfs ip
ips = sorted([ip for hostname, ip in cluster_ip.items()])
port = self.port # port='50070'
conn = None
for host in ips:
# 连接字符串
url = "http://{host}:{port}".format(host=host, port=port)
try:
conn = hdfs.Client(url, root='/', timeout=100, session=False)
conn.list('/')
break
except:
pass
# 如果执行到这一步，说明前面的都有问题
if conn.list('/'):
return conn
else:
raise Exception('没有找到hdfs的地址，请检查')
def __init__(self):
# 连接zookeeper
self.zk = KazooClient(hosts=self.zoo_quorum_dev)
self.zk.start()
# 判断是否存在zookeeper同步
if self.zk.exists(self.path_dev):
# 获取当前active的namenode
data = self.zk.get(self.path_dev)
ip = None
if self.name_node_a_dev[0] in data[0]:
ip = 'http://' + self.name_node_a_dev[1] + ':50070'
elif self.name_node_b_dev[0] in data[0]:
ip = 'http://' + self.name_node_b_dev[1] + ':50070'
if ip:
# 连接active的hdfs namenode
self.client = hdfs.Client(ip, root='/')
if not self.client:
print('hdfs client 建立连接失败!')
else:
print('nameNode IP 获取失败!')
def upload_file(hdfs_location, local):
try:
client = hdfs.Client('http://172.39.8.61:50070', root='/', timeout=10)
base_dir = local.split('/').pop() # 要上传的路径的最后一个文件夹
for root, dirs, files in os.walk(local):
new_dir = base_dir + root.split(base_dir).pop() # 去除本地路径前缀
for file in files:
old_path = root + '/' + file # 原始本地路径文件
lpath = new_dir + '/' + file # 去除本地路径前缀后的文件
if not client.status(hdfs_location + '/' + lpath,
strict=False):
# 第一个参数远程路径，第二个参数本地路径，第三个参数是否覆盖，第四个参数工作线程数
client.upload(hdfs_location + '/' + lpath,
old_path,
overwrite=False)
except Exception as e:
with open("err.log", "a") as f:
f.write(str(e))
f.write('\n')
def connect_hdfs(cls):
cls._client = hdfs.Client(**cls._parm_handler())
cls.handle_dns()
#dataframeoutput = dataframeoutput.drop('sum(EUCELL_DL_TPUT_NUM_KBITS)').drop('sum(DLPRBUSEDWITHDSPUC_FDUSERS)').drop('sum(DLPRBUSEDWITHDSPUC_FSUSERS)').drop('sum(EUCELL_DL_DRB_TPUT_NUM_KBITS)').drop('sum(EUCELL_DL_DRB_TPUT_DEN_SECS)')
dataframeoutput = dataframeoutput.select(
'DATE', 'MARKET', 'VENDOR', 'BAND', 'Cell Traffic (kbytes)',
'Cell Used PRB', 'Cell Spectral Efficiency (bps/Hz)',
'UE Traffic (kbytes)', 'UE Active Time (s)', 'UE Tput (kbps)',
'Total cell count', 'Total Spectrum in MHz')
dataframeoutput = dataframeoutput.coalesce(1)
#take action here
dataframeoutput.write.format('com.databricks.spark.csv').save(
outputName)
difference = dt.datetime.now() - start
dataframeoutput.unpersist()
sc.stop()
return difference
if __name__ == "__main__":
outDirectory = os.path.join(os.path.dirname(__file__), 'report/')
hdfsFiles = hdfs.Client('http://hdfs1:50070').list(
'/user/ec2-user/sample-data'
) # Use namenode public ip http://namenode:50070
print "start"
print ALU_LTE_SPARK().run("hdfs", hdfsFiles, outDirectory)
print "OK"
exit()
#Submit job to yarn on top of your hdfs cluster
#spark-submit --master yarn --deploy-mode cluster --num-executors 3 --executor-cores 2 --executor-memory 2g --packages com.databricks:spark-csv_2.10:1.5.0 spark-1.6-hdfs-yarn-CDH5.py
#For hdfs, pls find output file at:
#hdfs://user/ec2-user/report/result_group_by_MARKET_ALU_2017_spark_hdfs.csv
import hdfs
client = hdfs.Client('http://localhost:9870', root='/')
print(client.list('/'))
with client.write('/streaming.txt', encoding='utf-8') as hdfs_file:
for idx in range(10_000):
hdfs_file.write(f'{idx} {idx ** 2} {idx ** 3}\n')
hiddenLayer3 = add_layer("layer3", hiddenLayer2, in_size=512, out_size=128, activation_function=tf.nn.relu)
prediction = add_layer("end", hiddenLayer3, in_size=128, out_size=3, activation_function=tf.nn.softmax)
# loss = tf.reduce_mean(tf.reduce_sum(y_lable - prediction))
loss = -tf.reduce_mean(y_lable * tf.log(tf.clip_by_value(prediction, 1e-10, 1.0)))
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)
init = tf.global_variables_initializer()
sess = tf.Session()
sess.run(init)
client = hdfs.Client(HADOOP_IP_PORT, root="/", timeout=500, session=False)
fileList = client.list(HADOOP_PATH)
epoch_times = 0
while True:
epoch_times += 1
for file_loop in fileList: # 每个数据集中有一批数据
print('\n', file_loop)
get_data(client, HADOOP_PATH + file_loop)
# for i in range(2000):
#
def upload_pfile(self, cur_dir):
'''
构建字典存储爬虫插件对应的属性文件、原文文件夹、txt文本文件夹信息
键为插件名,值为属性文件位置、原文文件文件夹位置ori、txt文本文件夹位置txt组成的列表
当ori或txt文件夹不存在时,使用None占位
'''
f_all_dict = {}
print("mmmmmmmmmmmmmm", cur_dir)
for f in os.listdir(cur_dir):
if f.find('文献属性.xls') > 0 or f.find('文献属性.xlsx') > 0:
filepro = cur_dir + f if cur_dir[
-1] == '/' else cur_dir + '/' + f
filepath = cur_dir + f[:f.find('文献属性')] + '_ori/' if cur_dir[
-1] == '/' else cur_dir + '/' + f[:f.find('文献属性'
)] + '_ori/'
filetxt = cur_dir + f[:f.find('文献属性')] + '_txt/' if cur_dir[
-1] == '/' else cur_dir + '/' + f[:f.find('文献属性'
)] + '_txt/'
f_all_dict.setdefault(
f[:f.find('文献属性')], []).append(filepro) if os.path.exists(
filepro) else f_all_dict.setdefault(
f[:f.find('文献属性')], []).append(None)
f_all_dict.setdefault(
f[:f.find('文献属性')], []).append(filepath) if os.path.exists(
filepath) else f_all_dict.setdefault(
f[:f.find('文献属性')], []).append(None)
f_all_dict.setdefault(
f[:f.find('文献属性')], []).append(filetxt) if os.path.exists(
filetxt) else f_all_dict.setdefault(
f[:f.find('文献属性')], []).append(None)
for f_key in f_all_dict.keys():
if f_all_dict[f_key][0] is not None:
book = xlrd.open_workbook(f_all_dict[f_key][0])
sheet = book.sheet_by_index(0)
ut = time.strftime('%Y%m%d%H%M%S', time.localtime(time.time()))
ops = []
'''
逐行读取excel中信息,当爬取标注为CRA开头时,不添加上传标志UPA信息
'''
for r in range(1, sheet.nrows):
if sheet.cell(r, 0).value[0:3] == "CRA":
# values = (sheet.cell(r, 0).value[3:], sheet.cell(r, 2).value, sheet.cell(r, 3).value,sheet.cell(r, 4).value,sheet.cell(r, 5).value,sheet.cell(r, 6).value,sheet.cell(r, 7).value,sheet.cell(r, 8).value,sheet.cell(r, 9).value,sheet.cell(r,10).value,ut)
values = (sheet.cell(r,
0).value[3:], sheet.cell(r,
3).value,
sheet.cell(r, 4).value, sheet.cell(r,
5).value,
sheet.cell(r, 2).value, sheet.cell(r,
7).value,
sheet.cell(r, 8).value, sheet.cell(r,
9).value,
sheet.cell(r, 10).value, ut,
sheet.cell(r, 6).value)
# print("6666666666666", values)
ops.append(values)
else:
tag = "UPA" + time.strftime(
'%Y%m%d%H%M%S', time.localtime(time.time()))
values = (tag[3:], str(sheet.cell(r, 3).value),
str(sheet.cell(r, 4).value),
str(sheet.cell(r, 5).value),
tag + str(r).zfill(4) +
str(sheet.cell(r, 2).value),
str(sheet.cell(r, 7).value),
str(sheet.cell(r, 8).value),
str(sheet.cell(r, 9).value),
str(sheet.cell(r, 10).value), ut,
str(sheet.cell(r, 6).value))
# print("555555555555", values)
ops.append(values)
if f_all_dict[f_key][1] is not None:
f_list = os.listdir(f_all_dict[f_key][1])
f_list_doc = []
'''找到原文文件路径下所有文档,存到f_list_doc中'''
for f in f_list:
if os.path.splitext(f)[1] in {
'.caj', '.pdf', '.txt', '.doc', '.docx'
}:
self.suffix = os.path.splitext(f)[1]
f_list_doc.append(f)
temp_list = []
# print("f_list_doc", len(f_list_doc)) 40
for f in f_list_doc:
for item in ops:
# print("item", item[10][0:3])
if item[4][0:3] == 'CRA':
if item[4] == os.path.splitext(f)[0]:
temp_list.append(item)
else:
if item[4][21:] == os.path.splitext(f)[0]:
temp_list.append(item)
'''文件存在且与属性文件一一匹配的'''
save_file = []
for file in temp_list:
fileUuid = (str(uuid.uuid1()).replace("-", ""), )
if file[4][0:3] == 'CRA':
filepath = f_all_dict[f_key][1] + file[
4] + '.' + file[8]
else:
filepath = f_all_dict[f_key][1] + file[4][
21:] + '.' + file[8]
self.upload_filepath = filepath
try:
b = open(filepath, "rb").read()
origin = (pymysql.Binary(b), )
newfile = fileUuid + ("1010", "ZH") + file + origin
save_file.append(newfile)
except OSError:
if file[4][0:3] == 'CRA':
print('未找到文件' % file[4])
else:
print('未找到文件' % file[4][21:])
cursor = self.conn.cursor()
for n in range(0, len(save_file)):
a = globalVar.get_st()
print(a)
if a == 1:
self.CrawProcess.emit(
str("正在导入%s\n" % (save_file[n][7])))
try:
self.hdfs_ip = "http://192.168.1.107:50070"
self.inputpath = '/4516/upload'
self.client = hdfs.Client(self.hdfs_ip)
if self.configs['flag'] == True:
# cursor.executemany(
# "insert into DOCUMENTS(UUID,CRA_DT,TITLE,AUTHOR,AURDEPT,KYWRD,ABSTRACT,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,CONTENT_ORI,SOURCE_CODE,LANG)values(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s,%s,%s, %s, %s)",
# save_file[n:n+1])
# sql = "insert into DOCUMENTS(UUID,CRA_DT,TITLE,AUTHOR,AURDEPT,KYWRD,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,SOURCE_CODE,LANG,ABSTRACT,CONTENT_ORI)values(:1, to_date(:2,'yyyy-mm-dd hh24:mi:ss'), :3, :4, :5, :6, :7, :8, :9, :10, :11, :12, :13, :14, :15)"
# sql = "insert into DOCUMENTS(UUID,CRA_DT,TITLE,AUTHOR,AURDEPT,KYWRD,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,SOURCE_CODE,LANG,ABSTRACT)values(:1, to_date(:2,'yyyy-mm-dd hh24:mi:ss'), :3, :4, :5, :6, :7, :8, :9, :10, :11, :12, :13, :14)"
# txt导入
sql = "insert into DOCUMENTS(UUID,SOURCE_CODE,LANG,CRA_DT,AUTHOR,KYWRD,AURDEPT,TITLE,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,ABSTRACT,CONTENT_ORI)values(:1, :2, :3, to_date(:4,'yyyy-mm-dd hh24:mi:ss'), :5, :6, :7, :8, :9, to_date(:10,'yyyy-mm-dd hh24:mi:ss'), :11, :12, to_date(:13,'yyyy-mm-dd hh24:mi:ss'), :14,:15)"
# pdf导入
# sql = "insert into DOCUMENTS(UUID,SOURCE_CODE,LANG,CRA_DT,AUTHOR,KYWRD,AURDEPT,TITLE,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,ABSTRACT)values(:1, :2, :3, to_date(:4,'yyyy-mm-dd hh24:mi:ss'), :5, :6, :7, :8, :9, to_date(:10,'yyyy-mm-dd hh24:mi:ss'), :11, :12, to_date(:13,'yyyy-mm-dd hh24:mi:ss'), :14)"
# sql = "insert into DOCUMENTS(UUID,SOURCE_CODE,LANG,CRA_DT,AUTHOR,ABSTRACT,AURDEPT,KYWRD,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,TITLE)values('ce017578ec0411ea91d6a85e45b3a491', '1010', 'ZH', to_date('20200830203320','yyyy-mm-dd hh24:mi:ss'), '龙视要闻', '', '', '', '', to_date('20200830203320','yyyy-mm-dd hh24:mi:ss'), 'http://baijiahao.baidu.com/s?id=1676355714433432996', 'txt', to_date('20200830203320','yyyy-mm-dd hh24:mi:ss'), 'CRA202008302033200001美国最机密武器五年来首次现身，莫斯科：敢挑衅就摧毁')"
# sql = "insert into DOCUMENTS(UUID,SOURCE_CODE,LANG,CRA_DT,AUTHOR,ABSTRACT,AURDEPT,KYWRD,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,TITLE)values('fUiiiuid', '', '', '', '', '', '', '', '', '', '', '', '', '')"
# sql = "insert into DOCUMENTS(UUID,CRA_DT,TITLE,AUTHOR,AURDEPT,KYWRD)values('fUuid', to_date('2020-06-29 00:00:00','yyyy-mm-dd hh24:mi:ss'), 'hhhhhhhh', 'jjjjjjjjj', 'ooooo', 'ppppppp')"
# a = ('ce017578ec0411ea91d6a85e45b3a491', '1010', 'ZH', '20200830203320', '龙视要闻', '', '', '', '', '2020-08-29', 'http://baijiahao.baidu.com/s?id=1676355714433432996', 'txt', '20200901113956', 'CRA202008302033200001美国最机密武器五年来首次现身，莫斯科：敢挑衅就摧毁')
# 上传到oracle
cursor.executemany(
sql,
# # "insert into DOCUMENTS(UUID,CRA_DT,TITLE,AUTHOR,AURDEPT,KYWRD,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,SOURCE_CODE,LANG,ABSTRACT,CONTENT_ORI)values(:1, to_date(:2,'yyyy-mm-dd hh24:mi:ss'), :3, :4, :5, :6, :7, :8, :9, :10, :11, :12, :13, :14, :15)",
# # "insert into DOCUMENTS(UUID,CRA_DT,TITLE,AUTHOR,AURDEPT,KYWRD,ABSTRACT,JOURNAL,PUB_DT,URL,SUFFIX,UPLD_DT,CONTENT_ORI,SOURCE_CODE,LANG)values(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)",
# # "insert into DOCUMENTS(UUID,CRA_DT,TITLE)values(:1, :2, :3)",
save_file[n:n + 1])
# str(save_file[n:n+1]).replace('[','').replace(']',''))
# cursor.execute(sql)
try:
# 上传到hdfs
t = self.upload_filepath.rindex('/')
self.client.upload(
self.inputpath,
self.upload_filepath[0:t + 1] +
save_file[n][7] + self.suffix)
except Exception as e:
print("upload error!", e)
else:
break
except Exception as e:
print("1111111", e)
else:
break
self.CrawProcess.emit("导入完成")
cursor.close()
self.conn.commit()
if f_all_dict[f_key][2] is not None:
self.upload_txt(f_all_dict[f_key][2])
format='%(asctime)s - %(levelname)s: %(message)s')
cfg = configparser.ConfigParser()
cfg.read('config.conf')
cfg.sections()
hdfs_mvpath = cfg.get('path', 'hdfs_mvpath')
tmp_path = cfg.get('path', 'tmp_path')
oldhdfs_url = cfg.get('oldhdfs', 'url')
oldhdfs_root = cfg.get('oldhdfs', 'root')
newhdfs_url = cfg.get('newhdfs', 'url')
newhdfs_root = cfg.get('newhdfs', 'root')
dingrobot = cfg.get('alert', 'dingrobot')
notice = cfg.get('alert', 'notice')
oldhdfs = hdfs.Client(oldhdfs_url,
root=oldhdfs_root,
timeout=100,
session=False)
# newhdfs = hdfs.Client(newhdfs_url, root=newhdfs_root, timeout=100, session=False)
#oldhdfs = InsecureClient(oldhdfs_url, user="******")
newhdfs = InsecureClient(newhdfs_url, user="******")
L = threading.Lock()
hdfs_mvpathlist = hdfs_mvpath.strip(',').split(',')
class Producer(threading.Thread):
def __init__(self, name):
threading.Thread.__init__(self)
self.name = name
"""
存在GIL问题，所以能用进程的时候，用进程，或者用C语言解决
"""
hdfs_url = "http://192.168.100.120:50070"
hdfs_path = "/tmp/hive/hive"
last_day = 60 * 60 * 24 * 1000
# 删除前10天的文件
last_days = 9 * last_day
# 获取当前时间
current_time = round(time.time() * 1000)
# 创建hdfs连接
client = hdfs.Client(hdfs_url, timeout=100, session=False)
# 遍历删除文件所在目录下的文件，深度遍历
for first_depth in client.walk(hdfs_path, depth=1, status=True):
for second_depth in first_depth:
if type(second_depth) == list and len(second_depth) != 0:
# 输出所有的配置文件列表，里面以字典形式存放
file_list = second_depth
# 使用字典保存文件名和文件时间
list_files = []
for file_tuple in file_list:
# 生成文件名和文件修改最后修改时间的元组
dict_tuple = (file_tuple[0], file_tuple[1]["modificationTime"])
# 添加为列表
list_files.append(dict_tuple)
raise Exception(e)
finally:
cursor.close()
filepath = local_path + filename
with open(filepath, 'wb+') as f:
for row_data in res:
f.write(''.join(row_data).encode('utf-8') + '\n')
logger.info('写入本地文件%s完成' % filepath)
try:
logger.info('上传数据到hdfs')
txt_hdfs_path = hdfs_path + filename
client = hdfs.Client("http://192.10.86.31:50070",
root="/",
timeout=100,
session=False)
client.delete(hdfs_path, recursive=True)
client.upload(txt_hdfs_path, filepath)
logger.info('upload数据完成')
except Exception, e:
logger.error(e)
raise Exception(e)
oscmd = 'rm -f ' + filepath
logger.info('删除本地文件')
res = os.system(oscmd)
(retcode, retinfo) = ('success', '')
except Exception, e:
(retcode, retinfo) = ('fail', e)
for line in f.readlines():
tem = line[:-1].split(' ')
if len(tem)<2:
break
x = biao[tem[0]]
G[x]["label"] = tem[1]
node_num = len(G.keys())
neg = [0 for i in range(node_num)]
for i, node in enumerate(G.keys()):
neg[i] = negative[node]**0.75
s = sum(neg)
for i in range(1,node_num):
neg[i] = neg[i-1]+neg[i]
neg = [neg[i]/s for i in range(node_num)]
G['negative'] = neg
f.close()
f = open('graph.txt', 'w')
f.write(str(G))
f.close()
client = hdfs.Client("http://localhost:50070", timeout=100, session=False)
client.upload("/", "graph.txt", overwrite=True)
# with client.read('/graph.txt') as reader:
# G = eval(reader.read())
# a = list(G.keys())[:-1]
# print(a)
import hdfs
import pymongo
import json
import os
import time
# 启动 mongodb
# sudo mongod --dbpath=/Users/h2p/Documents/Project/data/db
client = hdfs.Client('http://*:50070', root='/')
print('连接 hdfs')
# client = hdfs.Client('http://*:50070', root='/')
# client = hdfs.Client('http://*:50070', root='/')
print('连接 mongodb')
# myClient = pymongo.MongoClient(host='*', port=20000)
myClient = pymongo.MongoClient(host='127.0.0.1', port=27017)
mydb = myClient['CloudComputing']
mycol = mydb['UserInfo']
print('读取已转移 Mongo Id')
Mongo_json_OK = []
with open('Mongo_json_OK.txt', 'r', encoding='utf-8') as f:
mongoId = f.readline().strip()
while mongoId:
Mongo_json_OK.append(id)
mongoId = f.readline().strip()
print('读取 Mongo 数据')
count = len(Mongo_json_OK)
for item in mycol.find():
df_fact.EUTRANCELLFDD == rdd_dimension.EUTRANCELLFDD,
'left').select(df_fact.EUTRANCELLFDD, 'DATETIME',
rdd_dimension.REGION, rdd_dimension.MARKET,
'PMACTIVEUEDLSUM')
outputDF.show()
outputDF.write.format('com.databricks.spark.csv').save(
'/Users/Joy4fun/Desktop/joined_' + filename)
#output to hdfs
#outputDF.write.format('com.databricks.spark.csv').save(outDirectory + filename)
print 'Writing is done!'
sc.stop()
if __name__ == "__main__":
intDirectory = "hdfs://hdfs2:8020/user/ec2-user/sample_data_eri/"
outDirectory = "hdfs://hdfs2:8020/user/ec2-user/joined_data/"
hdfs_fact_Files = hdfs.Client('http://hdfs2:50070').list(
'/user/ec2-user/sample_data_eri/')
hdfs_dimension_File = "hdfs://hdfs2:8020/user/ec2-user/ERI_CELL_REGION_MARKET.csv"
Spark_join_csv_in_hdfs("local", hdfs_fact_Files, hdfs_dimension_File,
hdfs_fact_Files, intDirectory, outDirectory)
#print Spark_read_write_csv_to_hdfs("s3", localFiles, outDirectory)
#spark-submit --packages com.databricks:spark-csv_2.10:1.5.0 Spark-1.6-read-write-CSV-to-hdfs-yarn-CDH5.py
#pyspark --packages com.databricks:spark-csv_2.11:1.5.0
#If you don't want to give --packages option, please:
# sudo cp downloads/spark-csv_2.11-1.5.0.jar /Library/Python/2.7/site-packages/pyspark/jars/.
import hdfs
import pymongo
import json
import os
import time
client = hdfs.Client('http://172.19.240.199:50070', root='/')
myClient = pymongo.MongoClient(host='172.19.240.199', port=20000)
mydb = myClient['CloudComputing']
mycol = mydb['UserInfo']
Mongo_json_OK = []
with open('Mongo_json_OK.txt', 'r', encoding='utf-8') as f:
mongoId = f.readline().strip()
while mongoId:
Mongo_json_OK.append(id)
mongoId = f.readline().strip()
count = len(Mongo_json_OK)
for item in mycol.find():
item['_id'] = str(item['_id'])
if item['_id'] not in Mongo_json_OK:
filePath = './json/' + item['_id'] + '.json'
with open(filePath, 'w', encoding='utf-8') as f:
json.dump(item, f, ensure_ascii=False)
client.upload('/streaminput/', filePath, overwrite=True)
os.remove(filePath)
Mongo_json_OK.append(item['_id'])
with open('Mongo_json_OK.txt', 'a', encoding='utf-8') as f:
def main(args):
server = tf.train.Server(cluster,
job_name=FLAGS.job,
task_index=FLAGS.task)
client = hdfs.Client("http://10.76.3.92:50070", root='/', timeout=100)
with client.read('/graph.txt') as reader:
G = eval(reader.read())
FLAGS.worker = len(cluster_dic['worker'])
FLAGS.node_num = len(G.keys()) - 1
# FLAGS.train_steps = FLAGS.node_num//FLAGS.batch_size*20
FLAGS.train_steps = 4
is_chief = (FLAGS.task == 0)
if FLAGS.job == 'ps':
server.join()
with tf.device(tf.train.replica_device_setter(cluster=cluster)):
global_step = tf.Variable(0, name='global_step',
trainable=False) # 创建纪录全局训练步数变量
emb_init = (np.random.randn(FLAGS.node_num, FLAGS.dim) /
np.sqrt(FLAGS.node_num / 2)).astype('float32')
emb = tf.Variable(emb_init, name='emb',
trainable=True) # 创建embedding向量并且初始化
L_con = 0
L_ucon = 0
# 训练参数占位符
pos = (FLAGS.w - 2 * FLAGS.cs) * 2 * FLAGS.cs
xc_0 = tf.placeholder(dtype=tf.int32,
shape=(pos * FLAGS.batch_size)) # 正边 source
xc_1 = tf.placeholder(dtype=tf.int32,
shape=(pos * FLAGS.batch_size)) # 正边 target
xuc_0 = tf.placeholder(dtype=tf.int32,
shape=(pos * FLAGS.ns *
FLAGS.batch_size)) # 无边 source
xuc_1 = tf.placeholder(dtype=tf.int32,
shape=(pos * FLAGS.ns *
FLAGS.batch_size)) # 无边 target
# 测试参数占位符
val = tf.placeholder(dtype=tf.int32,
shape=(2, FLAGS.test_size, 2)) # 随机抽取正边 = 负边
# 将边序列映射到embedding上
con_0_emb = tf.squeeze(tf.nn.embedding_lookup(emb,
xc_0)) # (batch, dim)
con_1_emb = tf.squeeze(tf.nn.embedding_lookup(emb,
xc_1)) # (batch, dim)
ucon_0_emb = tf.squeeze(tf.nn.embedding_lookup(emb,
xuc_0)) # (batch, dim)
ucon_1_emb = tf.squeeze(tf.nn.embedding_lookup(emb,
xuc_1)) # (batch, dim)
# 计算边相似度是，包括positive samples 和 negative samples
con_v = tf.sigmoid(
tf.squeeze(
tf.reduce_sum(tf.einsum('ni,ni->ni', con_0_emb, con_1_emb),
-1)))
ucon_v = tf.sigmoid(
tf.squeeze(
tf.reduce_sum(tf.einsum('ni,ni->ni', ucon_0_emb, ucon_1_emb),
-1)))
# 计算skip-gram的loss
L_con -= tf.reduce_sum(tf.log(con_v + 1e-15)) # connection
L_ucon -= tf.reduce_sum(tf.log(1 - ucon_v + 1e-15)) # unconnection
loss = (L_con + L_ucon)
optimizer = tf.train.AdamOptimizer(FLAGS.lr)
train_op = optimizer.minimize(loss, global_step=global_step)
AUC = tf.py_func(link_prediction, [val, emb], tf.double, stateful=True)
# init_op = tf.global_variables_initializer()# 参数初始化
with tf.train.MonitoredTrainingSession(
master=server.target,
is_chief=is_chief,
# hooks=[tf.train.StopAtStepHook(last_step=FLAGS.train_steps),
# tf.train.NanTensorHook(loss)],
# checkpoint_dir="./checkpoint_dir",
save_checkpoint_steps=100) as sess:
time_begin = time.time()
print('Traing begins @ %f' % time_begin)
local_step = 0
step = 0
dval = testing_data(FLAGS, G)
val_feed = {val: dval}
while not sess.should_stop() and step <= FLAGS.train_steps:
dxc_0, dxc_1, dxuc_0, dxuc_1 = traning_data(
FLAGS, G, local_step)
train_feed = {
xc_0: dxc_0,
xc_1: dxc_1,
xuc_0: dxuc_0,
xuc_1: dxuc_1
}
_, step, _loss = sess.run([train_op, global_step, loss],
feed_dict=train_feed)
local_step += 1
now = time.time()
print(
'%f: Worker %d: traing step %d dome (global step:%d/%d), and loss : %f'
% (now, FLAGS.task, local_step - 1, step,
FLAGS.train_steps, _loss))
if local_step % 10 == 0 and local_step != 0:
auc = sess.run([AUC], feed_dict=val_feed)
print("Link prediction AUC is %.2f" % auc[0])
if step >= FLAGS.train_steps:
break
auc = sess.run([AUC], feed_dict=val_feed)
print("Link prediction AUC is %.2f" % auc[0])
time_end = time.time()
print('Training ends @ %f' % time_end)
train_time = time_end - time_begin
print('Training elapsed time:%f s' % train_time)
sleep_time = 0
while sleep_time < 5:
time.sleep(2)
sleep_time += 1
print("Waiting other machines...")
import pyhdfs
import hdfs
if __name__ == '__main__':
# client = pyhdfs.HdfsClient(hosts='127.0.0.1:50070', user_name='Administrator')
# # client = client.get_active_namenode()
# client.mkdirs('/hadoop')
# print(client.listdir("/"))
# client.copy_from_local('E:/data/test/pku98/199801.txt', '/dfs/1.txt')
# print(client.listdir("/"))
client = hdfs.Client(url='http://127.0.0.1:50070')
client.makedirs('/test/hdfs', permission=777)
client.upload('/test/hdfs',
'E:/data/test/pku98/199801.txt',
overwrite=False)
def __client(path):
_, addr, file_path = path.split(':', 2)
client = hdfs.Client('http:' + addr + ':50070')
path = '/' + file_path.split('/', 1)[-1]
return client, path
#!/usr/bin/env python3
# coding=utf-8
# date 2019-05-15 17:23:25
# author calllivecn <*****@*****.**>
import hdfs
cli = hdfs.Client("http://192.168.56.6:9870")
for l in cli.list("/", True):
print(l)
