[文本处理] 批处理如何从html中提取出指定的几段内容？

@if(0)==(0) echo off& cscript.exe -nologo -e:jscript "%~f0"&pause&exit/b&@end

var aso = new ActiveXObject("ADODB.Stream");
aso.Mode = 3;
aso.Type = 2;
aso.Charset = "utf-8";
aso.Open();
aso.LoadFromFile("1.htm");
var txt = aso.ReadText(-1);
var reg = /\{"qid":.+?"answer":"([^"]+)"\}/g
var rs = txt.match(reg);
txt = "";
for(var i=0; i<rs.length; i++){
	reg = /\{"qid":(\d+),.+?(?="title)"title":"([^"&]+).*?","content":"(A、[^&]*).*?(B、[^&]*).*?(C、[^&]*).*?(D、[^\\]*).*?","optionnum":4,"answer":"([^"]+)"\}/g
	if(reg.test(rs[i]))
	txt+=RegExp.$1 				//qid
		+ "," + RegExp.$2 		//title
		+ "," + RegExp.$3 		//A
		+ "," + RegExp.$4 		//B
		+ "," + RegExp.$5 		//C
		+ "," + RegExp.$6 		//D
		+ "," + RegExp.$7		//answer
		+ "\r\n";
}
aso.Position = 0;
aso.WriteText(txt);
aso.SetEOS();
aso.SaveToFile("结果.txt", 2);
aso.Close();
aso = null;
复制代码

//&cls&dir /b "1.htm"|cscript.exe -nologo -e:jscript "%~f0"&pause&exit/b

var loadText = function (FileIn,Charset){
    var stm = new ActiveXObject('ADODB.Stream');
    stm.Type = 2;
    stm.Mode = 3;
    stm.Charset = Charset;
    stm.Open();
    stm.LoadFromFile(FileIn);
    stm.Position = 0;
    return stm.ReadText;
}
var saveText = function (str,FileOut,Charset){
    var stm = new ActiveXObject('ADODB.Stream');
    stm.Type = 2;
    stm.Mode = 3;
    stm.Charset = Charset;
    stm.Open();
    stm.WriteText(str);
    return stm.SaveToFile(FileOut,2);
}

while(!WScript.StdIn.AtEndOfStream) {
    var txt = '';
    var f= WScript.StdIn.ReadLine();
    var fn = f.replace(/\.[^\.]*$/,'');
    var txttmp = loadText(f,'utf-8').match(/\[\{"qid":.+?\}\]/)[0];
    var txttmp = txttmp.replace(/</g,'「').replace(/>/g,'」').replace(/「[^」]*」|\\\\r|\\\\n/g,'');
    var m = txttmp.match(/\{[^\}]+?\}/g);
    for (var i=0;i<m.length;i++){
        var qid = m[i].replace(/.+"qid":(\d+).+/,'$1');
        var title = m[i].replace(/.+?"title":"(.+)","content.+/,'$1');
        var content = m[i].replace(/.+?"content":"(.+)","optionnum.+/,'$1').replace(/\\n|\\t/g,'');
        var content = content.replace(/(B、|C、|D、)/g,'\r\n$1');
        var answer = '答案：'+m[i].replace(/.+?answer":"([^"]).+/,'$1');
        txt +=qid+'、'+title+'\r\n'+content+'\r\n'+answer+'\r\n\r\n'
    }
    saveText(txt,fn+'.txt','gb2312');
}
复制代码