python-tree磁盘大小查看-运维利器
经常磁盘告警，需要反复使用du -sh * 和 cd 来确认那些文件很大，需要删除或清理。
于是写了这个脚本，改吧改吧网上python实现的tree，对结果进行了排序
保留了原始的tree的功能，且对叶子文件夹大小排序输出，便于从后向前查看那些文件夹太大了~~
#! /usr/bin/env python
# -*- coding: gbk -*-
'''
@note : tree some path and display the leaf folder size by order
@authors : U{peterguo<mailto: [email protected]>}
@copyright: tencent
@date : 2013-02-25
@version : 1.0.0.1
'''
import os
import sys
from optparse import OptionParser
def dumplog(aStr, strMod="a"):
fp = file('run.log', strMod)
fp.write(str(aStr) + "\n")
fp.close()
def get_prefix(fs):
u'''get display tags according dir path
'''
s = ''
for i in range(0,len(fs)):
last = (i == len(fs)-1)
if fs[i] and last:
s = s + '|-'
elif fs[i] and not last:
s = s + '| '
elif not fs[i] and last:
s = s + '.-'
else:
s = s + ' '
if i != len(fs)-1:
s = s + ' '
else:
s = s + '-'
return s
def print_tree(current,fs,if_size,if_nofile):
u'''recurise display dirs
'''
global d_count,f_count,size,d_file2size,d_dir2size
fs.append(True)
dirs = []
try:
dirs = os.listdir(current)
except:
print "Exception on listdir [%s], ignore ......" % current
for i in range(0,len(dirs)):
if os.path.islink(current+os.sep+dirs[i]):
continue
abs_file = os.path.realpath(current+os.sep+dirs[i])
if i == len(dirs)-1:
fs[len(fs)-1] = False
if fs[len(fs)-1] and if_nofile:
last_flag = True
for j in range(i+1,len(dirs)):
if os.path.isdir(os.path.realpath(current+os.sep+dirs[j])):
last_flag = False
break
if last_flag:
fs[len(fs)-1] = False
if os.path.isdir(abs_file):
print '%s%s' % (get_prefix(fs),dirs[i])
dumplog('%s%s' % (get_prefix(fs),dirs[i]))
if not d_dir2size.has_key(abs_file):
d_dir2size[abs_file] = 0;
d_count = d_count + 1
print_tree(abs_file,fs[:],if_size,if_nofile)
else:
try:
file_size = os.path.getsize(abs_file)
d_file2size[abs_file] = file_size
if if_size:
size = size + file_size
if not if_nofile:
print '%s%s' % (get_prefix(fs),dirs[i])
dumplog('%s%s' % (get_prefix(fs),dirs[i]))
f_count = f_count + 1
except:
pass
def print_total(dir_count,file_count,if_size=False,file_size=0):
u'''print summary information
'''
if not if_size:
print u'total %d folder,%d files' % (d_count,f_count)
dumplog('total %d folder,%d files' % (d_count,f_count))
else:
unit = ('B','KB','MB','GB','TB')
i = len(str(size))/3
if i > 4:
i = 4
unit_size = 1024 ** i
f_size = round(float(size)/unit_size,2)
print u' total %d folders,%d files, size %s %s' % (d_count,f_count,f_size,unit[i])
dumplog(' total %d folders,%d files, size %s %s' % (d_count,f_count,f_size,unit[i]))
print ''
dumplog('')
parser = OptionParser()
parser.add_option("-p", "--paths", dest="paths", help="path")
parser.add_option("-s", "--size", dest="size", action="store_true", default=False, help="size the result")
parser.add_option("-f", "--nofile", dest="nofile", action="store_true", default=False, help="ignore file")
f_count = 0
d_count = 0
size = 0
d_dir2size = {} #record dir size
d_file2size = {} #record file size
if __name__ == "__main__":
options, args = parser.parse_args()
dumplog("start .......", 'wt')
if len(args)==0 :
print u'can not find target path, default as current path'
print os.getcwd()
d_dir2size[os.getcwd()] = 0
print_tree(os.getcwd(),[],options.size==True,options.nofile==True)
print_total(d_count,f_count,options.size==True,size)
else:
for path in args:
f_count = 0
d_count = 0
size = 0
if not os.path.exists(path):
print u'skip',path,u', can not find path.'
print ''
continue
path = os.path.realpath(path)
d_dir2size[path] = 0
print path
print_tree(os.path.normpath(path),[],options.size==True,options.nofile==True)
print_total(d_count,f_count,options.size==True,size)
#******************** summary every leaf folder *****************************#
for afile in d_file2size.keys():
d_dir2size[os.path.dirname(afile)] += d_file2size[afile]
l_out = []
for key in d_dir2size.keys():
if d_dir2size[key] != 0:
l_out.append("%10.4f MBytes \t%s" % (d_dir2size[key]/(1024*1024.0), key))
l_out.sort()
for item in l_out:
print item
dumplog(str(item))
输出举例：
python mytree.py -f /usr/local/apache2/ /usr/local/apache2 |--lib | .--pkgconfig |--bin |--build |--include |--modules |--conf | |--extra | |--original | | .--extra | .--soso.d |--htdocs |--error | .--include |--icons | .--small |--logs |--cgi-bin |--man | |--man1 | .--man8 .--manual |--developer |--faq |--howto |--images |--misc |--mod |--platform |--rewrite |--ssl |--style | |--css | |--lang | |--latex | .--xsl | .--util |--programs .--vhosts total 39 folder,1212 files 0.0000 MBytes /usr/local/apache2/htdocs 0.0004 MBytes /usr/local/apache2/conf/soso.d 0.0007 MBytes /usr/local/apache2/lib/pkgconfig 0.0010 MBytes /usr/local/apache2/cgi-bin 0.0013 MBytes /usr/local/apache2/error/include 0.0025 MBytes /usr/local/apache2/manual/style/latex 0.0125 MBytes /usr/local/apache2/manual/style 0.0127 MBytes /usr/local/apache2/conf/original 0.0140 MBytes /usr/local/apache2/manual/faq 0.0154 MBytes /usr/local/apache2/icons/small 0.0210 MBytes /usr/local/apache2/man/man1 0.0311 MBytes /usr/local/apache2/conf/extra 0.0311 MBytes /usr/local/apache2/conf/original/extra 0.0350 MBytes /usr/local/apache2/man/man8 0.0359 MBytes /usr/local/apache2/manual/style/css 0.0946 MBytes /usr/local/apache2/conf 0.1082 MBytes /usr/local/apache2/icons 0.1192 MBytes /usr/local/apache2/manual/images 0.1323 MBytes /usr/local/apache2/manual/rewrite 0.1482 MBytes /usr/local/apache2/error 0.1569 MBytes /usr/local/apache2/manual/developer 0.1682 MBytes /usr/local/apache2/manual/ssl 0.2167 MBytes /usr/local/apache2/manual/platform 0.2469 MBytes /usr/local/apache2/build 0.3739 MBytes /usr/local/apache2/manual/misc 0.3900 MBytes /usr/local/apache2/manual/howto 0.5180 MBytes /usr/local/apache2/manual/vhosts 0.5301 MBytes /usr/local/apache2/manual/programs 1.1219 MBytes /usr/local/apache2/include 2.3524 MBytes /usr/local/apache2/manual 2.6087 MBytes /usr/local/apache2/bin 3.1963 MBytes /usr/local/apache2/lib 5.3836 MBytes /usr/local/apache2/manual/mod 25.7610 MBytes /usr/local/apache2/modules 94.4569 MBytes /usr/local/apache2/logs
参考：http://qingyi.iteye.com/blog/1325347
附录：遍历子目录和子文件
def recursive_dir(strPath, strType="file"):
lRet = []
list_dirs = os.walk(strPath)
for root, dirs, files in list_dirs:
for d in dirs:
if strType in ("dir", "all"):
lRet.append(os.path.join(root, d))
for f in files:
if strType in ("file","all"):
lRet.append(os.path.join(root, f))
return lRet