Is that possible on linux?I have two or three lines which i need to change and it is contained in 240 files which is a lot of time requied to do with manual changing.First what i need to delete line which include cgi file,and second one is path which i need to change to proper path.

So how do i do that ?
Line which i want to delete is
<?php virtual('/cgi-bin/atx/in.cgi');?>

The -i option lets you edit the files in situ and will operate on each file instead of cat'ting the file arguments. You could use wildcard(s) instead of testfile2 or search for the lines in grep and pipe the filelist to the sed command using xargs.

Ok i will try that but i also finded one solution which works but it seems it requires to put slash before each character so i am little confused.
Here is working code:
perl -pi -w -e 's/BABA//g;' *.php
This replace all text called BABA to nothing.If i put
perl -pi -w -e 's/BABA/RABA/g;' *.php
all BABA text is replaced to RABA.
But it seems if you have special characters then you need to put slash,and that make entire thing little complicated.I finded that at this url:
http://www.liamdelahunty.com/tips/li...iple_files.php
I tried this one:
perl -pi -w -e 's/\<\?php\virtual\(\'\/cgi\-bin\/atx\/in\.cgi\'\)\;\?\>//g;' *.php
But it seems i missed something since it doesnt work.
Update:I finded what causing problem,it is this '
For some reason that blocks execution of command,probaly beacuse it have other meaning in that command.It seems i will have to find another solution for replacing text everytime when i encounter '
Update2:actualy i finded solution by simply replacing ' with " and it worked.This is how code should look:
perl -pi -w -e "s/\<\?php virtual\('\/cgi\-bin\/atx\/in\.cgi'\)\;\?\>//g;" *.php

My solution assumes that the line is by itself and deletes the line. Your perl example is similar to using sed:
sed 's/BABA//' filename

The reason for escaping those characters is because many of the characters in the line you gave are interpreted by the shell before it gets to the perl or sed command. I cheated a bit and wrote the sed command in a sed program. I still needed to escape the forward slash however.

If my assumption was wrong, then putting:
s#<?php virtual('/cgi-bin/atx/in.cgi');?>##
in a sed program will replace the matching string with nothing. To be more exacting, the . in "in.cgi" should be escaped as well. The "." is a single character wild card.
s#<?php virtual('/cgi-bin/atx/in\.cgi');?>##

Now if this is on a line by itself, there will be a blank line left over. Your perl example uses the same search/replace operation. The syntax is just a bit different. Sed is about 1/5 the size of perl, but if the rest of the script is a perl script, then use the perl example inside your script.

Your real decision is whether to learn perl or learn sed or both. The -i option does make using sed to process multiple files easier.

I used perl example and it works fine,just need to be careful with " and '
But now i encountered another problem,how to apply command to files in subfolders.
I tried adding -r or r to perl -pi -w -e 's/bla\/output/bla\/raba\_raba/g;' *.shtml
command but it doesnt work.Temporal solution will be to make batch command and putting it into usr bin so i can call it in each directory where i need to apply this search replace command but it will be better if i could simply type command only one time.

I guess there is no way to apply search & replace command to subdirectory files...

Use a loop of the type

There are many ways to do this. Instead of find you could try grep -Rl as in grep -Rl target-string parent-directory. The -R option searches directories recursively while the -l option simply lists the files that match the target string.

Alternatively you could do all of this in perl. Perl has a powerful built in find function which can search recursively for files with a certain extension and which match a certain string. You would then call a function which performs the actual text processing on that subset of files. You wouldn't even need a loop. A single recursive find call with appropriate filters would do the trick. What you were trying on the command line won't work but you could do anything you want in the actual script.

Well as you can see default command is
perl -pi -w -e 's/bla\/output/bla\/raba\_raba/g;' *.shtml
I am sure there is requied one more letter or something else to make it work but dont know what.

a while read loop would be better

............