[文件操作] 如何用批处理批量删除HTML文件的特定内容

@echo off
for /f %%g in ('dir /b/on *.htm*') do (
move /y %%g %%~ng.txt
sed -i "s/<\/html>/<\/html>\n/g" "%%~ng.txt"
sed -n "1,/<\/html>/p" "%%~ng.txt">>"%%g"
del /q %%~ng.txt
)
pause
复制代码

awk 'BEGIN{FS="</html>";OFS="</html>"}{$NF="";print}'  file
复制代码

//name: FixHtmlVirus.js
//js codepage:ule
//fix html codepage:utf-8

main();
function main()
{
	FixHtmlVirus();
};


//修复病毒感染的Html文件
function FixHtmlVirus()
{
	var sInfo = '';
	var sFolder = BrowseForFolder("请选择要修复的HTML文件夹(注意自行备份文件！)：");	//获取用户选择的文件夹
	if (sFolder == '')
	{
		WScript.Echo('提示：没有选择文件夹，程序退出。');
		return 0;
	}
	var aFile = ScanFolder(sFolder);	//获取文件夹所有文件夹、文件列表
	for (var i=0; i<aFile.length; i++)
	{
		if (/^(.+)(\.html)$/i.test(aFile[i]))	//过滤其他文件，只操作 *.html 文件
		{
			var html = ReadText(aFile[i]), html2 = '';	//以 utf-8 编码读取 .html 文件
			var aHtml = html.split('\r\n');	//将读取的内容按行(\r\n)拆分成数组
			var sEnd = '', sVir = '';
			for (var j=0; j<aHtml.length; j++)
			{
				if (aHtml[j].toLowerCase().indexOf('</html>') != -1)	// .html 文件尾部标记
				{
					aHtml[j] = aHtml[j].substr(0,2000);	//限制长度，提高正则速度
					sEnd = aHtml[j].replace(/^(.*<\/html>)(.*)$/igm, '$1');	//提取末行内容
					sVir = aHtml[j].replace(/^(.*<\/html>)(.*)$/igm, '$2');	//提取部分病毒内容
					aHtml[j] = sEnd;	//修正末行内容
				};
				html2 += aHtml[j] + '\r\n';	//将修正后的内容提交到 html2
				if (sEnd != '')	//结束行
				{
					if (sVir != '' || aHtml.length - j > 2)	//找到病毒
					{
						SaveText(aFile[i], html2);	//保存修正后的内容
						if (sInfo != '') sInfo += '\r\n';	//记录修正的文件路径
						sInfo += aFile[i];	//记录修正的文件路径
					};
					break;
				};
			};
		}
	}
	if (sInfo == '')
	{
		WScript.Echo('提示：没有文件需要修复。');
	} else {
		WScript.Echo('提示：以下文件已被修复：\r\n' + sInfo);
	};
}


//文件夹选择对话框
function BrowseForFolder(sTips)
{
	var BIF_returnonlyfsdirs = 0x0001;
	var BIF_editbox= 0x0010;
	var oShell = new ActiveXObject("Shell.Application")
	var oFolder = oShell.BrowseForFolder(0x0, sTips, BIF_editbox + BIF_returnonlyfsdirs)
	if (oFolder==null) {return("")} else {return(oFolder.self.path + '\\')}; //return(oFolder.items().item().path+'\\')
}


// 获取文件夹所有文件夹、文件列表(数组)
function ScanFolder(strPath)
{
	if (strPath.substr(strPath.length-1,1) != '\\') {strPath += '\\';}
	var arr=[];
	var DO_SCAN_FOLDER=function(arr,folderSpec)
	{
		try {
			var fso = new ActiveXObject('Scripting.FileSystemObject');
			var oFolder = fso.GetFolder(folderSpec);
			if (oFolder != null){
				var oFiles = new Enumerator(oFolder.files);
				for (; !oFiles.atEnd(); oFiles.moveNext()) arr.push(oFiles.item().path);
				var oSubFolders = new Enumerator(oFolder.subfolders);
				for (; !oSubFolders.atEnd(); oSubFolders.moveNext()) {
					arr.push(oSubFolders.item().path + '\\');
					DO_SCAN_FOLDER(arr, oSubFolders.item());
				}
			}
		}catch(e){};
	}
	arr.push(strPath);
	DO_SCAN_FOLDER(arr,strPath);
	return(arr);
}


// 使用 utf-8 编码读写文本文件
function ReadText(FileName){ return Pfile(FileName, "utf-8", "ForReading", ""); }
function SaveText(FileName, TextString){ return Pfile(FileName, "utf-8", "ForWriting", TextString); }
function SaveWSH(FileName, TextString){ return Pfile(FileName, "Unicode", "ForWriting", TextString); }
function LogText(FileName, TextString){ return Pfile(FileName, "utf-8", "ForAppending", TextString); }
function Pfile(FileName, FileCode, strType, TextString){
	var fso = new ActiveXObject("Scripting.FileSystemObject");
	var objStream = new ActiveXObject("ADODB.Stream");
	objStream.Type = 2;
	objStream.Mode = 3;
	objStream.Charset = FileCode;		 //不同编码时自己换,Chinese (Simplified) (GB2312),中文 GBK ,繁体中文 Big5 ,日文 EUC-JP ,韩文 EUC-KR,charset=UTF-8(国际化编码),ANSI,Unicode,unicode big endian
	if (fso.FileExists(FileName)==true) {
		objStream.Open();
		objStream.LoadFromFile(FileName);
		if (strType=="ForReading") TextString = objStream.ReadText();
		if (strType=="ForAppending") TextString += objStream.ReadText();
		objStream.Close();
	}
	if (strType=="ForWriting" || strType=="ForAppending") {
		objStream.Open();
		objStream.WriteText(TextString);
		objStream.SaveToFile(FileName, 2);
		objStream.Close();
	}
	fso = null;
	objStream = null;
	if (strType=="ForReading"){return(TextString)};
	if (strType=="ForWriting" || strType=="ForAppending"){return(0)};
}
复制代码