comments removed, I did not read the question correctly so the answer was not relevant.

Clearly sed can find the file, given that calling sed works fine as long as the regexp doesn't include ^ or \{2,\}.

I realize that I could manually open each file, read it into a variable, process it, and save it back to disk, but that seems cumbersome when I should be able to do it with one line with sed. The manual method also seems slower and a lot more memory intensive. I'll actually be processing thousands of files this way. But if you know of a more efficient way to do this from within Perl, I'm all ears.

You say "within" perl so I assume you are doing other things as well as what the sed command is doing. You can always use perls inplace editor, which is very fast. You can use it in a script or as a oneliner, this link should help you:

[perl.com...]

If you're running the perl script via cron or some other automated method you have to give full paths to file locations otherwise it won't run properly

JerryOdom, as I keep saying, my script can find the file just fine. That's not the problem.

perl_diver, thanks for the link. But I only see from that article how to run a regexp from the command line. I don't see how to do it from within a Perl script -- unless I'm supposed to open the file, read the contents, run the regexp, write the results, and close the file -- which defeats the purpose of my trying to do the substitution in one line. If you know how to do it simpler than that, could you post the syntax?

Sorry, I guess I got my reference materials confused. That link does not show how to use perls in place editor inside a script. This should help:

{ 
 local $^I = '.bak';  # emulates -i.bak 
 local @ARGV = glob("*.html"); # get list of .html files 
 while (<>) { 
  s/^ {2,}//; 
  print; 
 } continue {close ARGV if eof} 
} 

Thank you. I'm not sure that will work for me, though, because I'm using File::Find to recursively get a list of all files and process them one at a time, so I don't have any array of filenames. Further, I'm doing various tests on each file to decide whether I want to process it in the first place.

I guess my question should be: Given filename "$filename", what's the quickest/easiest way to run a search & replace on it from a Perl script?"

you can incorporate my code into your code.

You can use File::Slurp to efficiently read and write the entire file in one operation. You can then use normal Perl regular expressions to change the file.

To run any command from within perl you build the command in a string and pass that string to the system command.

IE:

$syscmd = "mv $filename2 $filename1";
system $syscmd;

will do a file move from within a perl script.

It should work with sed instead of mv.

Test it and let us know.