How to write a batch script in windows to loop over files , find a pattern and replace it

You can use Windows Scripting Host to get what you want.

Create a file called say, "1.wcf", and copy/paste the following:

<job>
    <script language="JavaScript">
        var fso = new ActiveXObject("Scripting.FileSystemObject");  
        var files = new Enumerator(fso.getFolder(".").files);
        var count = 0;
        for (; !files.atEnd(); files.moveNext())
        {
            var file = ""+files.item(); // make it string
            if (!file.match(/.*\.txt$/))
                { continue; WScript.echo("Found itself, skipping"); }
            //WScript.echo("Replacing in " + file);
            var f1 = fso.OpenTextFile(file, 1);
            var text = f1.ReadAll();
            f1.close();
            var lines = text.split("\r\n");
            for (var i = 0; i < lines.length; i++)
            {
              var m = lines[i].match(/^20(\d+)/);
              if (m)
              {
                lines[i] = lines[i].replace(new RegExp(m[1], "g"), '10101');
                //WScript.echo("Replaced in " + lines[i]);
              }
            }
            var f2 = fso.OpenTextFile(file, 2);
            f2.Write(lines.join("\r\n"));
            f2.close();
        }
        WScript.echo("Replaced "+count+" files");
    </script>
</job>

Then, copy this file into the folder with TXT files, and run. It will process each TXT, and if a line in the TXT file starts with 20, the rest of the adjoining digits are captured into Group 1, and then are used to replace all such digit sequences on that line.

Then, the file is re-written with the updated contents.

Your spec is a bit imprecise - position of 40 string is not as stated, and you don't state whether the spacing of the replacement line matters.

Since you tagged your question with regex, I think you will be interested in my JREPL.BAT regular expression text processing utility. It is pure script (hybrid JScript/batch) that runs natively on any Windows machine from XP onward.

This first solution simply replaces the digits following 20 and 40 with the new string, disregarding original string length. So the position of the 40 string may change (does change in your example).

@echo off
for /r %%F in (*) do call jrepl "^(1068 .*\n20)\d+( +40)\d+ *$" "$110101$210101" /m f "%%F" /o -

Here is a more complicated solution that preserves the position of the 40 string (position 332 in your example)

@echo off
for /r %%F in (*) do call jrepl "^(1068 .*\n20)(\d+ +)40\d+ *$" "$1+'10101'+Array($2.length-5+1).join(' ')+4010101" /m /j /f "%%F" /o -

This final solution assumes the line is formatted with fixed width, and both the 20 and 40 numbers have maximum length of 10. This solution preserves both the position of the numbers, and the total length of the line:

@echo off
for /r %%F in (*) do jrepl jrepl "^(1068 .*\n20)\d+ *( {322}40)\d+ *$" "$110101   $210101   " /m /f "%%F" /o -

The method below assume that there are not empty lines in the files. This point may be fixed, if needed.

@echo off
setlocal EnableDelayedExpansion

rem Set working values
set "find=20"
set "replace=10101"

rem Process all files in current folder and below it
for /F "delims=" %%a in ('dir /A-D /S /B *.*') do (

   rem Read this file via redirected input
   rem and create a .tmp extension copy of it via redirected output
   < "%%a" (

      rem Read the first line
      set /P "line="
      set lastLine=1

      rem Find the number of the lines that start with "20"
      for /F "tokens=1,2 delims=: " %%b in ('findstr /N "^%find%" "%%a"') do (

         rem Copy the lines before this one
         set /A lines=%%b-lastLine, lastLine=%%b
         for /L %%i in (1,1,!lines!) do set /P "line=!line!" & echo/

         rem Process this line as desired:
         rem Get the first token in this line
         set "token=%%c"
         rem Get the pattern to replace removing "20" from beginning of the token
         rem and replace it in the entire line
         for /F %%d in ("!token:*%find%=!") do set "line=!line:%%d=%replace%!"

      )

      rem Copy the last replaced line
      echo !line!

      rem Copy the rest of lines after the last replaced one
      findstr "^"

   ) > "%%~Na.tmp"

   rem Replace the original file by the processed one
   move /Y "%%~Na.tmp" "%%a" > NUL

)