去除C++注释的脚本

'
' 去除CPP注释.vbs
' 2006/11/16
' Yi Wang
'
' Ver3.0.0
' 能够区分""中的注释符号。除了多行字符串:
' "a multi-line /
' String literal signals its /
' continuation with a backslash."
'
' 用法：
' 将需要处理的文件托拽到该脚本上。
'
' 注意：
' 使用脚本之前请先备份源代码文件!!!
' 脚本长时间没有响应，到任务管理器中关闭WScript.exe进程，结束脚本执行。
' 作者不为使用脚本造成的任何损失负责。
'
' 修改记录：
' 2.0.2 (1) 修改处理"/*//"类似代码时，搜索起点的BUG。
'       (2) 报告错误信息时添加文件名。
' 3.0.0 (1) 添加递归文件夹的功能
'       (2) 添加识别文件后缀名的功能，能识别的后缀名：H,C,CPP,CXX,INC,JAVA
'       (3) 报告结果由没处理一个文件之后，改到所有都处理完毕之后。
'       (4) 报告结果写入同时生成的日志中。
'-----------------------------------------------

' Global
Const VBSNAME = "去除CPP注释"
Const SHORT_NAME = "CCC"
Dim NOWFILE
RESULT_REPORT = ""
LOGFILE_NAME = SHORT_NAME & " 处理结果记录.txt"  '日志文件,文件将被放到桌面上

' Usage
usage = "该脚本用于去除文件中C++代码风格的注释。" & vbCr
usage = usage & "将需要处理的文件托拽到该脚本上。" & vbCr & vbCr
usage = usage & "* 使用脚本之前请先备份源代码文件！！！" & vbCr
usage = usage & "* 作者不为使用脚本造成的任何损失负责。"

' SHELL
Set wshshell = CreateObject( "WScript.Shell" )
LOGFILE_NAME = wshshell.specialFolders("desktop") & "/" & LOGFILE_NAME

' File System Interface
Const ForReading = 1, ForWriting = 2, ForAppending = 8
Set fs = CreateObject( "Scripting.FileSystemObject" )
TMPPATH = wshshell.ExpandEnvironmentStrings( "%TMP%" )
TMPFILE = TMPPATH & "/vbstmp.tmp"

' Arguments
Set args = WScript.Arguments
If args.count <= 0 Then
    MsgBox usage, vbInformation, VBSNAME
    WScript.Quit
Else
    ' 处理参数的中的文件
    For Each path In args
        If fs.FolderExists( path ) Then
            Set fd = fs.GetFolder( path )
            CleanFolder( fd )
        Else
            NOWFILE = path
            CleanComment NOWFILE
        End If
    Next
End If

' 收尾工作
If fs.FileExists( TMPFILE ) Then
    fs.DeleteFile( TMPFILE )
End If

' 报告处理结果
MsgBox "处理完毕", vbInformation, VBSNAME
If Not fs.FileExists( LOGFILE_NAME ) Then
    fs.createTextFile( LOGFILE_NAME )
End If
Set logfile = fs.OpenTextFile( LOGFILE_NAME, ForWriting )
logfile.write RESULT_REPORT
logfile.close
wshshell.run "notepad " & LOGFILE_NAME

'*****************************
' Sub CleanFolder
Sub CleanFolder( srcFolder )
    For Each file In srcFolder.Files
        NOWFILE = file.path
        CleanComment NOWFILE
    Next
    For Each fd In srcFolder.subFolders
        CleanFolder fd
    Next
End Sub

'*****************************
' Sub CleanComment
Sub CleanComment ( path )
    '##. 文件类型检查
    If Not CheckExtName( fs.GetExtensionName( path ) ) Then
        Exit Sub
    End If

    report = ""

    '##. 检查该文件是否存在，不存在exit sub
    If Not fs.FileExists(path) Then
        report = path & " 文件不存在！"
        MsgBox report, vbCritical, VBSNAME
        Exit Sub
    End If

    '##. 源文件拷贝到 TMPFILE
    fs.CopyFile path, TMPFILE

    '##. 打开文件
    Set srcfile = fs.OpenTextFile( TMPFILE, ForReading )
    Set desfile = fs.OpenTextFile( path, ForWriting )

    '##. 删除注释
    lineCnt = 0
    codeline = ""
    bInBlock = False
    Do While srcfile.AtEndOfStream <> True
        ' 读取
        codeline = srcfile.ReadLine
        ' 以行为单位处理代码
        CleanCodeLine codeline, bInBlock
        ' 写入
        desfile.WriteLine codeline
        lineCnt = lineCnt + 1
    Loop

    '##. 关闭文件
    srcfile.close
    desfile.close

    '## 通报结果
    report = path & vbCrLf & vbTab & "处理完毕。" & vbCrLf
    report = report & vbTab & "扫描 " & lineCnt & " 行。" & vbCrLf
    RESULT_REPORT = RESULT_REPORT & report
End Sub

'**********************************
' Sub CleanCodeLine
Sub CleanCodeLine( ByRef codeline, ByRef bInBlock )
    ' block comment pointer
    pta = 0
    pta2 = 0
    ' line comment pointer
    ptb = 0
    ' comma pointer
    ptc = 0

    ' buffer line
    bufline = ""

    If bInBlock Then
        ' 在注释块/**/中
        pta2 = InStr( codeline, "*/" )
        If pta2 <= 0 Then
            codeline = ""
        Else
            codeline = Right( codeline, Len(codeline) - pta2 - 1 )
            bInBlock = False
            CleanCodeLine codeline, bInBlock
        End If
    Else
        ' 不在注释块中
        pta = InStr( codeline, "/*" )
        ptb = InStr( codeline, "//" )
        ptc = InStr( codeline, """" )
        pta2 = InStr( codeline, "*/" )

        If pta2 > 0 And pta2 < pta Then
            ' 代码错误 /* */ 不匹配
            MsgBox NOWFILE & vbCr & "代码错误 /* */ 不匹配！脚本中止！", vbCritical, VBSNAME
            WScript.Quit
        End If

        Do While ( pta <> 0 ) Or ( ptb <> 0 )
            ' 存在注释
            bSwCh = False
            If ptc > 1 Then
                If Mid(codeline, ptc-1, 1) = "/" Then
                    bSwCh = True
                End If
            End If
            bCommented = False
            If ptc > 0 Then
                If ( pta > 0 And pta < ptc ) Or ( ptb > 0 And ptb < ptc ) Then
                    bCommented = True
                End If
            End If
            If ( ptc > 0 ) And ( Not bSwCh ) And ( Not bCommented ) Then
                ' 存在字符串
                bufline = bufline & Left( codeline, ptc )
                codeline = Right( codeline, Len(codeline) - ptc )
                strlen = Len( codeline )
                For i = 1 To strlen
                    bSwCh = False
                    If i > 1 Then
                        If Mid( codeline, i-1, 1 ) = "/" Then
                            bSwCh = True
                        End If
                    End If
                    If ( Mid( codeline, i, 1 ) = """" ) And ( Not bSwCh ) Then
                        Exit For
                    End If
                Next
                If i > strlen Then
                    ' 该行没找到下一个字符串界定符号"，代码有误
                    MsgBox NOWFILE & vbCr & "字符串界定符号不匹配！脚本中止！", vbCritical, VBSNAME
                    WScript.Quit
                Else
                    ' 将字符串的内容放入buffer line
                    ptc = i
                    bufline = bufline & Left( codeline, ptc )
                    codeline = Right( codeline, Len(codeline) - ptc )
                End If
                pta = InStr( codeline, "/*" )
                ptb = InStr( codeline, "//" )
                ptc = InStr( codeline, """" )
            ElseIf ( pta = 0 ) Or ( ptb <> 0 And pta > ptb ) Then
                codeline = Left( codeline, ptb - 1 )
                Exit Do
            Else
                pta2 = InStr( pta+2, codeline, "*/" )
                If pta2 <= 0 Then
                    bInBlock = True
                    codeline = Left( codeline, pta - 1 )
                    Exit Do
                Else
                    If pta > pta2 Then
                        ' 代码错误 /* */ 不匹配
                        MsgBox NOWFILE & vbCr & "代码错误 /* */ 不匹配！脚本中止！", vbCritical, VBSNAME
                        WScript.Quit
                    End If
                    codeline = Left( codeline, pta - 1) & Right( codeline, Len(codeline) - pta2 - 1 )
                    pta = InStr( codeline, "/*" )
                    ptb = InStr( codeline, "//" )
                    ptc = InStr( codeline, """" )
                End If
            End If
        Loop
        codeline = bufline & codeline

        pta2 = InStr( codeline, "*/" )
        If pta2 > 0 Then
            ' 代码错误 /* */ 不匹配
            MsgBox NOWFILE & vbCr & "代码错误 /* */ 不匹配！脚本中止！", vbCritical, VBSNAME
            WScript.Quit
        End If
    End If
End Sub

'*******************************
' Function CleanLineComment
Function CleanLineComment( codeline )
    CleanLineComment = Left( codeline, InStr( codeline, "//" ) - 1 )
End Function

'*******************************
' Function CheckExtName
' 能识别的后缀名：H,C,CPP,CXX,INC,JAVA
Function CheckExtName( extname )
    extname = UCase( extname )
    If extname = "H" Then
        CheckExtName = True
    ElseIf extname = "C" Then
        CheckExtName = True
    ElseIf extname = "CPP" Then
        CheckExtName = True
    ElseIf extname = "CXX" Then
        CheckExtName = True
    ElseIf extname = "INC" Then
        CheckExtName = True
    ElseIf extname = "JAVA" Then
        CheckExtName = True
    Else
        CheckExtName = False
    End If
End Function

 

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

用来测试用的cpp文件：

#include <stdio.h>

int main(int argc, char *argv[])
{
    printf("Hello, world/n");

    [1]-------------------------------------------------------
    block comment - a line
    /*block comment - a line*/

    [2]-------------------------------------------------------
    block comment - multilines
    /*line1
    line2
    line3*/

    [3]-------------------------------------------------------
    block comment - nest itself
    drawString( g/*Graphics*/, x/*x-coordinate*/, y/*yy*/, 0 );/*ssssdijfoas
    ndzxc*/

    [4]-------------------------------------------------------
    line comment
    setences();//line comment
    //line comment2

    [5]-------------------------------------------------------
    comment nested.
    nested1;//cccccc/*--
    nested2;//---*/
    nested3;/*cccccccccccc // line comment symbol
    ccccccccc//ccccccccccc*/
    nested4./*ccccccc*/Part1;/*ccccc // line comment symbol
    ccccccccc//ccccccccccc*/
    nested5./*ccccccc*/Part1./*ccccc*/Part2;// line comment symbol

    [6]--------------------------------------------------------
    String s1 = "//not lien comment";
    String s2 = "/*not block comment*/";
    String s3 = "http://www.url.net";// "http://www.urlOld.net";

    return 0;
}