动态向Pandas数据框添加数据
我最近在玩Pandas，想把HTTP日志导入Pandas进行分析，因为这是一种很好的大数据源，也能让我学习Pandas。
我每次只能接收一行日志，所以不能直接从CSV文件导入，需要把这些日志“灌输”到Pandas的DataFrame中，然后再保存到HDFStore文件里。
目前我写的代码是从GZIP文件读取日志，这样我可以先启动这个过程，但等我把Pandas的部分搞定后，我会把它改成事件驱动的方式，类似于发布-订阅的协程。
这是我目前的代码：
import os
import gzip, io
import pandas as pd
import numpy as np
def read_gzip(file):
if os.path.isdir(os.path.dirname(file)):
if os.path.isfile(file):
while True:
for line in io.BufferedReader(gzip.open(file)):
yield line
class ElbParser(object):
def __init__(self, file):
self.file = file
def get_log_lines(self):
return read_gzip(self.file)
def build_series(self, log_line):
def normalise_ip(input_string):
try:
text = input_string.split(':')[0]
except:
text = np.NaN
finally:
return text
def normalise_time(input_string):
try:
text = float(input_string)
except:
text = np.NaN
finally:
return text
def normalise_ints(input_string):
try:
text = int(input_string)
except:
text = np.NaN
finally:
return text
log_list = log_line.split()
elb_series = pd.Series({
'timestamp' : np.datetime64(log_list[0]),
'elb_name' : log_list[1],
'client_ip' : normalise_ip(log_list[2]),
'backend_ip' : normalise_ip(log_list[3]),
'request_processing_time' : normalise_time(log_list[4]),
'backend_processing_time' : normalise_time(log_list[5]),
'response_processing_time' : normalise_time(log_list[6]),
'elb_status_code' : normalise_ints(log_list[7]),
'backend_status_code' : normalise_ints(log_list[8]),
'received_bytes' : normalise_ints(log_list[9]),
'sent_bytes' : normalise_ints(log_list[10]),
'http_method' : log_list[11].strip("'").strip('"'),
'request' : log_list[13].strip("'").strip('"')
})
return elb_series
下面是用来测试上述代码的基本测试代码：
import os
import sys
import unittest
import pandas as pd
from lbreader import ElbParser
test_elb_log_file = 'tests/resources/lb_log.gz'
class ElbTest(unittest.TestCase):
#
# Fixture Framework
#
def setUp(self):
print '\nTest Fixture setUp'
self.elbparser = ElbParser(test_elb_log_file)
def tearDown(self):
print '\nTest Fixture tearDown'
del self.elbparser
#
# Tests
#
def test_01_get_aws_elb_log_line(self):
print '\n running get aws elb log line'
log_lines = self.elbparser.get_log_lines()
self.failUnlessEqual('2014-05-15T23:00:04.696671Z poundlb 192.168.0.10:53345 - -1 -1 -1 408 408 55 0 "PUT https://localhost:443/private/urls HTTP/1.1"\n'
, log_lines.next())
def test_02_build_series_with_aws_elb_log_line(self):
print '\n running build series from elb log'
log_lines = self.elbparser.get_log_lines()
test = self.elbparser.build_series(log_lines.next())
test_line_dict = {'backend_ip': '-',
'backend_processing_time': -1.0,
'backend_status_code': 408,
'client_ip': '192.168.0.10',
'lb_name': 'poundlb',
'elb_status_code': 408,
'http_method': 'PUT',
'received_bytes': 55,
'request': 'HTTP/1.1',
'request_processing_time': -1.0,
'response_processing_time': -1.0,
'sent_bytes': 0,
'timestamp': numpy.datetime64('2014-05-16T00:00:04.696671+0100')}
self.failUnlessEqual(test_line_dict, test.to_dict())
if __name__ == '__main__':
unittest.main()
现在我遇到的问题是：
我能创建pandas.Series数据，但当我尝试把它放进DataFrame时，它却变成了两列11行的数据。
test.values
>>> array(['-', -1.0, 408, '192.168.0.10', 'poundlb', 408, 'PUT', 55,
'HTTP/1.1', -1.0, -1.0, 0,
numpy.datetime64('2014-05-16T00:00:04.696671+0100')], dtype=object)
df = pd.DataFrame(test.values.tolist(), column_list)
>>> df
Out[35]:
0
backend_ip -
backend_processing_time -1
backend_status_code 408
client_ip 192.168.0.10
elb_name poundlb
elb_status_code 408
http_method PUT
received_bytes 55
request HTTP/1.1
request_processing_time -1
response_processing_time -1
sent_bytes 0
timestamp 2014-05-16T00:00:04.696671+0100
[13 rows x 1 columns]
这和我希望得到的结果相差甚远，我原本希望得到的是[1行 x 1列]：
backend_ip backend_processing_time backend_status_code backend_status_code client_ip elb_name elb_status_code http_method received_bytes request request_processing_time response_processing_time sent_bytes timestamp
0 - -1 408 408 192.168.0.10 poundlb 408 PUT 55 HTTP/1.1 -1 -1 0 2014-05-16T00:00:04
1 - -1 408 408 192.168.0.11 poundlb 408 PUT 55 HTTP/1.1 -1 -1 0 2014-05-16T00:00:05
2 - -1 408 408 192.168.0.12 poundlb 408 PUT 55 HTTP/1.1 -1 -1 0 2014-05-16T00:00:06
这样如果我再添加一行pandas.Series日志，就能得到另一行，依此类推。我还打算在以下字段上建立索引：时间戳、客户端IP、后端IP。
我真的很希望能得到一些帮助，因为我似乎搞不清楚我的行和列是怎么回事。
经过一番尝试，我得到了以下结果：所以在玩了一会儿之后，我得到了这个结果，但仍然无法连接或追加。
>>> test0 = elbparser.build_series(log_lines.next())
>>> test1 = elbparser.build_series(log_lines.next())
>>> test2 = elbparser.build_series(log_lines.next())
>>> test3 = elbparser.build_series(log_lines.next())
>>> test4 = elbparser.build_series(log_lines.next())
>>> test5 = elbparser.build_series(log_lines.next())
>>> test6 = elbparser.build_series(log_lines.next())
>>> test7 = elbparser.build_series(log_lines.next())
>>> test8 = elbparser.build_series(log_lines.next())
>>> test9 = elbparser.build_series(log_lines.next())
>>> test10 = elbparser.build_series(log_lines.next())
>>> test_list = [test.to_dict(), test1.to_dict(), test2.to_dict(), test3.to_dict(), test4.to_dict(), test5.to_dict(), test6.to_dict(), test7.to_dict(), test8.to_dict(), test9.to_dict(), test10.to_dict()]
>>> test_list
[{'backend_ip': '-',
'backend_processing_time': -1.0,
'backend_status_code': 408,
'client_ip': '192.168.0.1',
'elb_name': 'poundlb',
'elb_status_code': 408,
'http_method': 'PUT',
'received_bytes': 55,
'request': 'HTTP/1.1',
'request_processing_time': -1.0,
'response_processing_time': -1.0,
'sent_bytes': 0,
'timestamp': numpy.datetime64('2014-05-16T00:00:04.696671+0100')},
{'backend_ip': '10.0.0.241',
'backend_processing_time': 59.246736,
'backend_status_code': 403,
'client_ip': '192.168.0.2',
'elb_name': 'poundlb',
'elb_status_code': 403,
'http_method': 'PUT',
'received_bytes': 55,
'request': 'HTTP/1.1',
'request_processing_time': 3.4e-05,
'response_processing_time': 2.9e-05,
'sent_bytes': 64,
'timestamp': numpy.datetime64('2014-05-16T00:00:30.980494+0100')},
{'backend_ip': '10.0.0.242',
'backend_processing_time': 59.42053,
'backend_status_code': 200,
'client_ip': '192.168.0.3',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'PUT',
'received_bytes': 173,
'request': 'HTTP/1.1',
'request_processing_time': 8.4e-05,
'response_processing_time': 6.9e-05,
'sent_bytes': 149,
'timestamp': numpy.datetime64('2014-05-16T00:00:32.687835+0100')},
{'backend_ip': '10.0.0.39',
'backend_processing_time': 0.016443,
'backend_status_code': 200,
'client_ip': '192.168.0.4',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'GET',
'received_bytes': 0,
'request': 'HTTP/1.1',
'request_processing_time': 7.9e-05,
'response_processing_time': 4e-05,
'sent_bytes': 289,
'timestamp': numpy.datetime64('2014-05-16T00:00:38.247760+0100')},
{'backend_ip': '10.0.0.41',
'backend_processing_time': 0.008624,
'backend_status_code': 200,
'client_ip': '192.168.0.5',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'GET',
'received_bytes': 0,
'request': 'HTTP/1.1',
'request_processing_time': 5.4e-05,
'response_processing_time': 3.2e-05,
'sent_bytes': 200,
'timestamp': numpy.datetime64('2014-05-16T00:00:38.432535+0100')},
{'backend_ip': '10.0.0.43',
'backend_processing_time': 0.138925,
'backend_status_code': 200,
'client_ip': '192.168.0.6',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'GET',
'received_bytes': 0,
'request': 'HTTP/1.1',
'request_processing_time': 4.5e-05,
'response_processing_time': 4.3e-05,
'sent_bytes': 268,
'timestamp': numpy.datetime64('2014-05-16T00:00:38.509598+0100')},
{'backend_ip': '10.0.0.38',
'backend_processing_time': 0.013578,
'backend_status_code': 200,
'client_ip': '192.168.0.7',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'POST',
'received_bytes': 291,
'request': 'HTTP/1.1',
'request_processing_time': 4.2e-05,
'response_processing_time': 2.7e-05,
'sent_bytes': 36,
'timestamp': numpy.datetime64('2014-05-16T00:00:38.667479+0100')},
{'backend_ip': '10.0.0.42',
'backend_processing_time': 0.017493,
'backend_status_code': 200,
'client_ip': '192.168.0.8',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'GET',
'received_bytes': 0,
'request': 'HTTP/1.1',
'request_processing_time': 3.7e-05,
'response_processing_time': 2.7e-05,
'sent_bytes': 290,
'timestamp': numpy.datetime64('2014-05-16T00:00:38.708697+0100')},
{'backend_ip': '10.0.0.40',
'backend_processing_time': 0.014167,
'backend_status_code': 200,
'client_ip': '192.168.0.9',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'POST',
'received_bytes': 297,
'request': 'HTTP/1.1',
'request_processing_time': 3.5e-05,
'response_processing_time': 2.7e-05,
'sent_bytes': 36,
'timestamp': numpy.datetime64('2014-05-16T00:00:38.746867+0100')},
{'backend_ip': '10.0.0.40',
'backend_processing_time': 0.094383,
'backend_status_code': 200,
'client_ip': '192.168.0.10',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'PUT',
'received_bytes': 79,
'request': 'HTTP/1.1',
'request_processing_time': 3.4e-05,
'response_processing_time': 3.6e-05,
'sent_bytes': 148,
'timestamp': numpy.datetime64('2014-05-16T00:00:39.333482+0100')},
{'backend_ip': '10.0.0.42',
'backend_processing_time': 0.061355,
'backend_status_code': 200,
'client_ip': '192.168.0,10',
'elb_name': 'poundlb',
'elb_status_code': 200,
'http_method': 'PUT',
'received_bytes': 79,
'request': 'HTTP/1.1',
'request_processing_time': 9.6e-05,
'response_processing_time': 5.8e-05,
'sent_bytes': 148,
'timestamp': numpy.datetime64('2014-05-16T00:00:39.345097+0100')}]
>>> df = pd.DataFrame(test.to_dict(), index_list)
>>> df
Out[45]:
backend_ip backend_processing_time
backend_ip - -1
backend_processing_time - -1
backend_status_code - -1
client_ip - -1
elb_name - -1
elb_status_code - -1
http_method - -1
received_bytes - -1
request - -1
request_processing_time - -1
response_processing_time - -1
sent_bytes - -1
timestamp - -1
backend_status_code client_ip
backend_ip 408 192.168.0.10
backend_processing_time 408 192.168.0.10
backend_status_code 408 192.168.0.10
client_ip 408 192.168.0.10
elb_name 408 192.168.0.10
elb_status_code 408 192.168.0.10
http_method 408 192.168.0.10
received_bytes 408 192.168.0.10
request 408 192.168.0.10
request_processing_time 408 192.168.0.10
response_processing_time 408 192.168.0.10
sent_bytes 408 192.168.0.10
timestamp 408 192.168.0.10
elb_name elb_status_code http_method
backend_ip poundlb 408 PUT
backend_processing_time poundlb 408 PUT
backend_status_code poundlb 408 PUT
client_ip poundlb 408 PUT
elb_name poundlb 408 PUT
elb_status_code poundlb 408 PUT
http_method poundlb 408 PUT
received_bytes poundlb 408 PUT
request poundlb 408 PUT
request_processing_time poundlb 408 PUT
response_processing_time poundlb 408 PUT
sent_bytes poundlb 408 PUT
timestamp poundlb 408 PUT
received_bytes request request_processing_time
backend_ip 55 HTTP/1.1 -1
backend_processing_time 55 HTTP/1.1 -1
backend_status_code 55 HTTP/1.1 -1
client_ip 55 HTTP/1.1 -1
elb_name 55 HTTP/1.1 -1
elb_status_code 55 HTTP/1.1 -1
http_method 55 HTTP/1.1 -1
received_bytes 55 HTTP/1.1 -1
request 55 HTTP/1.1 -1
request_processing_time 55 HTTP/1.1 -1
response_processing_time 55 HTTP/1.1 -1
sent_bytes 55 HTTP/1.1 -1
timestamp 55 HTTP/1.1 -1
response_processing_time sent_bytes
backend_ip -1 0
backend_processing_time -1 0
backend_status_code -1 0
client_ip -1 0
elb_name -1 0
elb_status_code -1 0
http_method -1 0
received_bytes -1 0
request -1 0
request_processing_time -1 0
response_processing_time -1 0
sent_bytes -1 0
timestamp -1 0
timestamp
backend_ip 2014-05-15 23:00:04.696671
backend_processing_time 2014-05-15 23:00:04.696671
backend_status_code 2014-05-15 23:00:04.696671
client_ip 2014-05-15 23:00:04.696671
elb_name 2014-05-15 23:00:04.696671
elb_status_code 2014-05-15 23:00:04.696671
http_method 2014-05-15 23:00:04.696671
received_bytes 2014-05-15 23:00:04.696671
request 2014-05-15 23:00:04.696671
request_processing_time 2014-05-15 23:00:04.696671
response_processing_time 2014-05-15 23:00:04.696671
sent_bytes 2014-05-15 23:00:04.696671
timestamp 2014-05-15 23:00:04.696671
[13 rows x 13 columns]
这是我想要的结果，但在这之后我似乎还是有问题，无法进行追加或连接。
我会继续研究解决方案。
1 个回答
1
这段代码是用来处理一些数据的。它的主要目的是从一个地方获取信息，然后对这些信息进行一些操作，最后再把结果返回给你。具体来说，它可能会从数据库中提取数据，或者从用户输入中获取信息。
在这段代码中，首先会定义一些变量，这些变量就像是用来存储信息的盒子。接着，代码会执行一些逻辑，比如判断某个条件是否成立。如果条件成立，代码就会执行特定的操作；如果不成立，可能会执行其他的操作。
最后，代码会将处理后的结果返回给调用它的地方，这样你就可以看到最终的输出了。
总的来说，这段代码的功能就是获取、处理和返回数据，帮助你完成特定的任务。
import pandas as pd
import numpy
# Create a data dict
test_line_dict = {'backend_ip': '-',
'backend_processing_time': -1.0,
'backend_status_code': 408,
'client_ip': '192.168.0.10',
'lb_name': 'poundlb',
'elb_status_code': 408,
'http_method': 'PUT',
'received_bytes': 55,
'request': 'HTTP/1.1',
'request_processing_time': -1.0,
'response_processing_time': -1.0,
'sent_bytes': 0,
'timestamp': numpy.datetime64('2014-05-16T00:00:04.696671+0100')}
# Create the corresponding dataframe
idOfTheFirstElement = '1st'
df=pd.DataFrame.from_dict({idOfTheFirstElement:test_line_dict})
# Transpose the result
df = df.T
# One can dynamically append/add new rows with pd.concat
dfNew = pd.concat([df,df,df])