tr/// is the transliteration operator, good for converting individual characters into some other character (or counting, or deleting individual characters). It's not helpful in this case.

You actually want to look at the s/// substitution operator, for what you're trying to accomplish. Here is a description of what to do:

open $filename for reading. Open a temporary file for output. Scan through $filename line by line using the while( <INFILE> ) {..... construct. Inside the loop, on each line, use the substitution operator to identify matches, and substitute for your new text. Then print each line back out to your temp file. After the loop terminates, close both file handles, and rename the temp file to replace the original file.

If you're operating in a multi-user or web server context, file locking will also be advisable.

perl -i.orig -pe 'BEGIN{$localvar = "xyz"}; s!http://domain.com/folder
+/subfolder!http://domain.com/newfolder/newsubfolder/$localvar!g' your
+_filename more_files
[download]

What I do is I have a script in my ~/bin directory called fixfile.pl that looks like this:

!/usr/bin/perl -p -i

s/expression_to_replace/replacement_expression/gs;
[download]

My program is much longer, because each time I find a regex that works I leave it in the program, commented, and add my new regex. That way I create a library of useful regexes. To use the program I just go to the directory I want to make the changes in, and run (for your example) fixfile.pl *.html and the regex is performed on all files with the extension .html. The reason I find this useful is because I can leverage the power of a perl one-liner without losing the record of what I've been doing, and it lets me use my familiar shell globbing operators.

I think the script you are trying to write can be wrote with Tie::File . Read the documentation about it.