统计某文件夹下源代码行数 Python脚本

写了一个统计某文件夹下各种源代码行数的脚本

仅支持c/c++ java python

有需要可以自己更改程序，我把这个脚本的拓展性做的很高

很容易使之支持其他文本型语言代码

或者给我Email我帮你做 onezeros@yahoo.cn

对于MFC自动生成的代码应该排除在外

可以用一种不太严密的方法：先统计各种mfc生成工程代码数量

然后在分析文件夹时根据mfc文件的一些特征判断目标文件夹是否为mfc工程

如果是，就减去生成的代码量

#!/usr/bin/python
#codelines.py short for files in directory
import sys
import os
import os.path

class Element:
_fileamount=0#number of files
_codelines=0 #line number of files
_commandlines=0;#standalone command lines

extension={'c':0,'cpp':0,'C':0,'cc':0,'cxx':0,
'h':0,'hpp':0,'hxx':0,'java':1,'py':2}
language=['c/c++','java','py']
statistics=[Element() for x in range(0,len(language))]

#count lines in a file
def fileparser(file,ext):
global extension,statistics
try:
f=open(file,'r')
except:
print("failed to open "+file)
print("parsing file "+file)
#print("extension "+str(extension[ext]))
statistics[extension[ext]]._fileamount+=1
try:
context=f.readlines()
i=-1;
while i<len(context):
i+=1
line=context[i].strip()
if line=='':
continue
else:
if extension[ext]==0 or ext=='java':#c/cpp#java
if line[0:2]=='//':
statistics[extension[ext]]._commandlines+=1
elif line[0:2]=='/*':
while line.find('*/')==-1:
i+=1;
line=context[i].strip()
statistics[extension[ext]]._commandlines+=1
statistics[extension[ext]]._commandlines+=1
else:
statistics[extension[ext]]._codelines+=1
elif ext=='py':#python
if line[0]=='#':
statistics[extension[ext]]._commandlines+=1
else:
statistics[extension[ext]]._codelines+=1
except:
pass
f.close()
return
#find source code file by analyzing extension
def extparser(file):
global extension
ext=os.path.splitext(file)[1][1:]
#print("extension "+ext)
if ext in extension:
#print("in  extparing  "+file)
fileparser(file,ext)
else:
pass
return

#parse a directory to find files
def dirparser(directory):
print("parsing directory "+directory)
try:
dirlist=os.listdir(directory)
for i in dirlist:
fullpath=directory+'//'+i
if(os.path.isdir(fullpath)==True):
dirparser(fullpath)
else:
extparser(fullpath)
except:
pass
return

if __name__ == "__main__":
if(os.path.isdir(sys.argv[1])==False):
print('%s is not a correct path',sys.argv[1]);
sys.exit(0)
dirparser(sys.argv[1])
length=[len('language:  '),len('file amount: '),
len('code amount: '),len('command amount')]
print('language:  file amount:  code amount:  command amount')
for i,lang in enumerate(language):
print(repr(lang).ljust(length[0]),
repr(statistics[i]._fileamount).ljust(length[1]),
repr(statistics[i]._codelines).ljust(length[1]),
repr(statistics[i]._commandlines).ljust(length[1]))

print('Done')

下图是我的一个文件夹的统计结果