Python处理utf-8 添加和删除BOM头

以下代码只处理了assic和utf8文件。其它文件编码为保险起见并未加入支持。

参数

exts 需要处理文件的扩展名

folders 需要处理的文件夹及子目录

处理目录为当前目录

运行：

添加bom头

python proc_bom.py

删除bom头

python proc_bom.py -r

运行缺少chardet报错 请查看：http://www.cnblogs.com/wangmh/p/8004451.html

1 #!/usr/bin/python
 2 # -*- coding: UTF-8 -*-
 3 
 4 import os;
 5 import sys;
 6 import codecs;
 7 import chardet;
 8 
 9 #获取脚本文件的当前路径
10 def cur_file_dir():
11      #获取脚本路径
12      path = sys.path[0]
13      #判断为脚本文件还是py2exe编译后的文件，如果是脚本文件，则返回的是脚本的目录，如果是py2exe编译后的文件，则返回的是编译后的文件路径
14      if os.path.isdir(path):
15          return path
16      elif os.path.isfile(path):
17          return os.path.dirname(path)
18 #打印结果
19 
20 
21 #pip install chardet 安装相应插件
22 def procBOM(strPath,curLen, bAdd):
23     newcontent = '';
24     f = open(strPath, "rb");
25     fcontent = f.read();
26     f.close();
27     printBuffer = strPath[curLen:]
28     codeType = chardet.detect(fcontent)["encoding"]  #检测编码方式  
29      printBuffer = printBuffer + "  "+str(codeType) 
30 
31      if codeType.lower().find('utf-8') == -1 and codeType.lower().find('ascii') == -1 :
32          #非utf8文件保险起见先退出,并输出错误提示,todo后续再添加其它转码到utf8
33          print printBuffer + " error OK"
34          return
35     
36     #不需要转换，已经添加bom头    
37 
38     if bAdd and fcontent[:3] != codecs.BOM_UTF8:
39         print  printBuffer+" add bom",
40         newcontent = codecs.BOM_UTF8;
41         newcontent += fcontent;
42     elif not bAdd and fcontent[:3] == codecs.BOM_UTF8:
43         newcontent = fcontent[3:];
44         print  printBuffer+" del bom",
45     else:
46         return;
47     fnew = open(strPath, "wb+")
48     fnew.write(newcontent);
49     fnew.close();
50     print "done"
51     return
52 
53 if __name__ == "__main__":
54 
55     bAdd = True;
56     exts = ['.h', '.c', '.cpp'];
57     folders = ["GNaviInterface/search","src","tester"]
58     bAdd = True;
59     if(len(sys.argv) > 1 and sys.argv[1] == '-r'):
60         bAdd = False;
61     curLen = len(cur_file_dir())
62     for folderName in folders:
63         folderPath = cur_file_dir()+"/"+folderName+"/"
64         #print "procBOM:folder path = "+folderPath+",add = "+str(bAdd)
65         for parent,dirnames,filenames in os.walk(folderPath):
66             for f in filenames:
67                 bTargetFile = False;
68                 for e in exts:
69                     if(f.endswith(e)):
70                         bTargetFile = True;
71                 if(bTargetFile):
72                     procBOM(os.path.join(parent,f),curLen, bAdd);
73 #print 'file:%s add:%s' % (os.path.join(parent, f), bAdd);      

转载于:https://www.cnblogs.com/wangmh/p/8005388.html