使用Python去除C/C++源程序中的所有注释和空行

说明

使用Python的字符串处理和正则表达式处理实现了一个删除C/C++源程序中所有注释和空行的小脚本。

使用字符串处理

# delete all the comments and empty line of a C/C++ source file
import os, sys,string

#-------------------------------------------------------------
def usage():
    print u'''
    help: del_comment.py <filename | dirname>
    '''
#--------------------------------------------------------------
def deal_file(src):
    # file exist or not
    if not os.path.exists(src):
        print 'Error: file - %s doesn\'t exist.'% src
        return False
    if os.path.islink(src):
        print 'Error: file - %s is a link.'
        return False
    filetype = (os.path.splitext(src))[1]
    if not filetype in ['.c','.h','.cpp','.hh','.cc']:
        return False
    try:
        if not os.access(src, os.W_OK):
            os.chmod(src, 0664)
    except:
        print 'Error: you can not chang %s\'s mode.'% src

    inputf = open(src, 'r')
    outputfilename = (os.path.splitext(src))[0] + '_no_comment'+filetype
    outputf = open(outputfilename, 'w')

    try:
        #-----find /*.....*/
        rFlag=0
        line=inputf.readline()
        while(line):
            fm=string.find(line,'/*')
            if fm!=-1:  # find a /*
                if fm>1:# not empty line
                    outputf.write(line[:fm]+'\n')
                rFlag=1
                # find */
                fm=string.find(line,'*/')
                if fm!=-1:
                    rFlag=0
                else:
                    line=inputf.readline()
                    while line:
                        fm=string.find(line,'*/')
                        if fm!=-1:
                            rFlag=0
                            break
                        line=inputf.readline()
                    if not line:
                        print 'Match /*...*/ error'
            else: # deal with //
                fm=string.find(line,'//')
                if fm==-1:
                    if len(line)>1: # not empty line
                        outputf.write(line)
                elif fm!=-1 and (not rFlag):
                    if fm>1: # not empty line
                        outputf.write(line[:fm]+'\n')
            #read nextline
            line=inputf.readline()
    except:
        print 'Error: unexcept error.'
        inputf.close()
        outputf.close()
    return True

#--------------------------------------------------------------
def deal_dir(src):
    #  dir exist or not
    if not os.path.exists(src):
        print 'Error: dir - %s is not exist.'%s (src)
        return False
    filelists = os.listdir(src)
    for eachfile in filelists:
        eachfile = src + '/' +eachfile
        if os.path.isdir(eachfile):
            deal_dir(eachfile)
        elif os.path.isfile(eachfile):
            deal_file(eachfile)
    return True

#--------------------------------------------------------------
def main():
    if len(sys.argv) < 2:
        usage()
        sys.exit(1)
    src = sys.argv[1]
    # get absolute dir/file path
    if os.path.isdir(src):
        dire = os.path.abspath(src)
        dirFlag = True
    elif os.path.isfile(src):
        fl = os.path.abspath(src)
        dirFlag = False
    else:
        print 'File input error'

    # deal
    if dirFlag:
        deal_dir(dire)
    else:
        deal_file(fl)
    print 'Successful handle file.'

#--------------------------------------------------------------
if __name__ == '__main__':
    main()

使用正则表达式

# delete all the comments and empty line of a C/C++ source file
import os, sys,string,re,glob

# /*..*/  //...
Rule1 = "(\/\*(\s|.)*?\*\/)|(\/\/.*)"
c1=re.compile(Rule1)

#-------------------------------------------------------------
def usage():
    print u'''
    help: del_comment.py <filename | dirname>
    '''
#--------------------------------------------------------------
def deal_file(src):
    # file exist or not
    if not os.path.exists(src):
        print 'Error: file - %s doesn\'t exist.'% src
        return False
    if os.path.islink(src):
        print 'Error: file - %s is a link.'
        return False
    filetype = (os.path.splitext(src))[1]
    if not filetype in ['.c','.h','.cpp','.hh','.cc']:
        return False
    try:
        if not os.access(src, os.W_OK):
            os.chmod(src, 0664)
    except:
        print 'Error: you can not chang %s\'s mode.'% src

    inputf = open(src, 'r')
    outputfilename = (os.path.splitext(src))[0] + '_no_comment'+filetype
    outputf = open(outputfilename, 'w')

    lines=inputf.read()
    inputf.close()
    lines=re.sub(Rule1,"",lines)
    outputf.write(lines)
    outputf.close()
    return True

#--------------------------------------------------------------
def deal_dir(src):
    #  dir exist or not
    if not os.path.exists(src):
        print 'Error: dir - %s is not exist.'%s (src)
        return False
    filelists = os.listdir(src)
    for eachfile in filelists:
        eachfile = src + '/' +eachfile
        if os.path.isdir(eachfile):
            deal_dir(eachfile)
        elif os.path.isfile(eachfile):
            deal_file(eachfile)
    return True

#--------------------------------------------------------------
def main():
    if len(sys.argv) < 2:
        usage()
        sys.exit(1)
    src = sys.argv[1]
    # get absolute dir/file path
    if os.path.isdir(src):
        dire = os.path.abspath(src)
        dirFlag = True
    elif os.path.isfile(src):
        fl = os.path.abspath(src)
        dirFlag = False
    else:
        print 'File input error'

    # deal
    if dirFlag:
        deal_dir(dire)
    else:
        deal_file(fl)
    print 'Successful handle file.'

#--------------------------------------------------------------
if __name__ == '__main__':
    main()

使用示例

待处理文件：

#ifndef _RS232_H_
#define _RS232_H_

/* the maximum number of ports we are willing to open */
#define MAX_PORTS 4

/*this array hold information about each port we have opened */
struct PortInfo{
    int busy;
    char name[32];
    int handle;
};

int OpenCom(int portNo,const char deviceName[],long baudRate);
int CloseCom(int portNo);
int ComRd(int portNo,char buf[],int maxCnt,int Timeout);
int ComWrt(int portNo,const char * buf,int maxCnt);

//long GetBaudRate(long baudRate);
//int OpenComConfig(int port,
//                  const char deviceName[],
//                  long baudRate,
//                  int parity,
//                  int dataBits,
//                  int stopBits,
//                  int iqSize,
//                  int oqSize);

#endif

处理结果：

#ifndef _RS232_H_
#define _RS232_H_
#define MAX_PORTS 4
struct PortInfo{
    int busy;
    char name[32];
    int handle;
};
int OpenCom(int portNo,const char deviceName[],long baudRate);
int CloseCom(int portNo);
int ComRd(int portNo,char buf[],int maxCnt,int Timeout);
int ComWrt(int portNo,const char * buf,int maxCnt);
#endif

在使用Python正则表达式处理的时候，有一个问题，就是没法删除文件中的空行。如果想删除空行，可能还得一行行读进来，把长度为0的行不保存。这是目前感觉到的难点。