#!/usr/bin/env python
# coding=utf-8
#——————————————————
# name: apache 日志分析脚本
# purpose: 此脚本只用来分析apache的访问日志
# version: 2.0
# author: leo
# created: 2013-4-26
# modified: 2013-5-4
# copyright: (c) leo 2013
#——————————————————
import sys
import time
#该类是用来打印格式
class displayformat(object):
def format_size(self,size):
”’格式化流量单位”’
kb = 1024
mb = 1048576
gb = 1073741824
tb = 1099511627776
if size >= tb :
size = str(size / tb) + ‘t’
elif size < kb :
size = str(size) + 'b'
elif size >= gb and size < tb:
size = str(size / gb) + 'g'
elif size >= mb and size < gb :
size = str(size / mb) + 'm'
else :
size = str(size / kb) + 'k'
return size
formatstring = '%-15s %-10s %-12s %8s %10s %10s %10s %10s %10s %10s %10s'
def transverse_line(self) :
'''输出横线'''
print self.formatstring % ('-'*15,'-'*10,'-'*12,'-'*12,'-'*10,'-'*10,'-'*10,'-'*10,'-'*10,'-'*10,'-'*10)
def head(self):
'''输出头部信息'''
print self.formatstring % ('ip','traffic','times','times%','200','404','500','403','302','304','503')
def error_print(self) :
'''输出错误信息'''
print
print 'usage : ' + sys.argv[0] + ' apachelogfilepath [number]'
print
sys.exit(1)
def execut_time(self):
'''输出脚本执行的时间'''
print
print "script execution time: %.3f second" % time.clock()
print
#该类是用来生成主机信息的字典
class hostinfo(object):
host_info = ['200','404','500','302','304','503','403','times','size']
def __init__(self,host):
self.host = host = {}.fromkeys(self.host_info,0)
def increment(self,status_times_size,is_size):
'''该方法是用来给host_info中的各个值加1'''
if status_times_size == 'times':
self.host['times'] += 1
elif is_size:
self.host['size'] = self.host['size'] + status_times_size
else:
self.host[status_times_size] += 1
def get_value(self,value):
'''该方法是取到各个主机信息中对应的值'''
return self.host[value]
#该类是用来分析文件
class fileanalysis(object):
def __init__(self):
'''初始化一个空字典'''
self.report_dict = {}
self.total_request_times,self.total_traffic,self.total_200,
self.total_404,self.total_500,self.total_403,self.total_302,
self.total_304,self.total_503 = 0,0,0,0,0,0,0,0,0
def split_eachline_todict(self,line):
'''分割文件中的每一行，并返回一个字典'''
split_line = line.split()
split_dict = {'remote_host':split_line[0],'status':split_line[-2],'bytes_sent':split_line[-1],}
return split_dict
def generate_log_report(self,logfile):
'''读取文件，分析split_eachline_todict方法生成的字典'''
for line in logfile:
try:
line_dict = self.split_eachline_todict(line)
host = line_dict['remote_host']
status = line_dict['status']
except valueerror :
continue
except indexerror :
continue
if host not in self.report_dict :
host_info_obj = hostinfo(host)
self.report_dict[host] = host_info_obj
else :
host_info_obj = self.report_dict[host]
host_info_obj.increment('times',false)
if status in host_info_obj.host_info :
host_info_obj.increment(status,false)
try:
bytes_sent = int(line_dict['bytes_sent'])
except valueerror:
bytes_sent = 0
host_info_obj.increment(bytes_sent,true)
return self.report_dict
def return_sorted_list(self,true_dict):
'''计算各个状态次数、流量总量，请求的总次数，并且计算各个状态的总量 并生成一个正真的字典，方便排序'''
for host_key in true_dict :
host_value = true_dict[host_key]
times = host_value.get_value('times')
self.total_request_times = self.total_request_times + times
size = host_value.get_value('size')
self.total_traffic = self.total_traffic + size
o200 = host_value.get_value('200')
o404 = host_value.get_value('404')
o500 = host_value.get_value('500')
o403 = host_value.get_value('403')
o302 = host_value.get_value('302')
o304 = host_value.get_value('304')
o503 = host_value.get_value('503')
true_dict[host_key] = {'200':o200,'404':o404,'500':o500,'403':o403,'302':o302,'304':o304,
'503':o503,'times':times,'size':size}
self.total_200 = self.total_200 + o200
self.total_404 = self.total_404 + o404
self.total_500 = self.total_500 + o500
self.total_302 = self.total_302 + o302
self.total_304 = self.total_304 + o304
self.total_503 = self.total_503 + o503
sorted_list = sorted(true_dict.items(),key=lambda t:(t[1]['times'],t[1]['size']),reverse=true)
return sorted_list
class main(object):
def main(self) :
'''主调函数'''
display_format = displayformat()
arg_length = len(sys.argv)
if arg_length == 1 :
display_format.error_print()
elif arg_length == 2 or arg_length == 3:
infile_name = sys.argv[1]
try :
infile = open(infile_name,'r')
if arg_length == 3 :
lines = int(sys.argv[2])
else :
lines = 0
except ioerror,e :
print
print e
display_format.error_print()
except valueerror :
print
print "please enter a volid number !!"
display_format.error_print()
else :
display_format.error_print()
fileanalysis_obj = fileanalysis()
not_true_dict = fileanalysis_obj.generate_log_report(infile)
log_report = fileanalysis_obj.return_sorted_list(not_true_dict)
total_ip = len(log_report)
if lines :
log_report = log_report[0:lines]
infile.close()
print
total_traffic = display_format.format_size(fileanalysis_obj.total_traffic)
total_request_times = fileanalysis_obj.total_request_times
print 'total ip: %s total traffic: %s total request times: %d'
% (total_ip,total_traffic,total_request_times)
print
display_format.head()
display_format.transverse_line()
for host in log_report :
times = host[1]['times']
times_percent = (float(times) / float(fileanalysis_obj.total_request_times)) * 100
print display_format.formatstring % (host[0],
display_format.format_size(host[1]['size']),
times,str(times_percent)[0:5],
host[1]['200'],host[1]['404'],
host[1]['500'],host[1]['403'],
host[1]['302'],host[1]['304'],host[1]['503'])
if (not lines) or total_ip == lines :
display_format.transverse_line()
print display_format.formatstring % (total_ip,total_traffic,
total_request_times,'100%',
fileanalysis_obj.total_200,
fileanalysis_obj.total_404,
fileanalysis_obj.total_500,
fileanalysis_obj.total_403,
fileanalysis_obj.total_302,
fileanalysis_obj.total_304,
fileanalysis_obj.total_503)
display_format.execut_time()
if __name__ == '__main__':
main_obj = main()
main_obj.main()