I'm not sure that there is a 'p' modifier for regular expressions (did you mean to use the '-p' command-line switch?) and there is no point in using the 'x' modifier if you don't use extended syntax. This can be done in a one-liner like this.

$ mkdir lloderl74
$ cat > lloderl74/script1
#!/usr/bin/perl -w
print "Hello World!\n";
exit;
$ cat > lloderl74/script2
#!/usr/bin/perl -w
die "Goodbye, Cruel World\n";
$ perl -pi.bak -e '
s{^(#!/usr/bin/perl.*)}{$1\n\nuse English q{no_match_vars}}' lloderl74
+/*
$ head -99 lloderl74/*
==> lloderl74/script1 <==
#!/usr/bin/perl -w

use English q{no_match_vars}
print "Hello World!\n";
exit;

==> lloderl74/script1.bak <==
#!/usr/bin/perl -w
print "Hello World!\n";
exit;

==> lloderl74/script2 <==
#!/usr/bin/perl -w

use English q{no_match_vars}
die "Goodbye, Cruel World\n";

==> lloderl74/script2.bak <==
#!/usr/bin/perl -w
die "Goodbye, Cruel World\n";
[download]

I hope this helps you.

Thanks, johngg. This unfortunately does not help because of my wanting to avoid the command line and I also want to adhere to Perl Best Practices, (PBP)! I have changed the $INPUT_RECORD_SEPARATOR to undefined on purpose so that I may slurp in an entire file at a time, because this code is a small part of a script and this is more efficient than rewriting the file a line at a time. If you'll notice, I am using the English module within this code and I originally used ${^MATCH}, (which requires the /p expression modifier), instead of $1 but it also did not work! Help still needed! P.S. My apologies for a mistakenly created subject line which does not really help!

You shouldn't need to use PBP for one-liners - that defeats the point. And if you're so worried about it, shouldn't you also be using strict and warnings? In that case, you would see the compilation error

Bareword "no_match_vars" not allowed while "strict subs" in use at ...

promptly fix it with use English qw( -no_match_vars ) and be on your merry way. :)

Update - added the qw (thanks moritz).

junior application developer

lloder174: You can go back and Update the title of your post (making a note that you have done so), and you can at the same time fix the problem that your code does not have a valid  </code> end tag, which renders it almost unreadable.

johngg: I believe the  //p regex modifier is new with 5.10: see p in perlre.

Your specific problem (things being printed in the wrong order) appears to be due to the combination of double quotes and the x modifier. The double quotes in my $searchPattern = "(^\#!/usr/bin/perl.*?\n)"; causes "\#" to be inserted in the string as plain "#". The x modifier causes Perl to read "#" as the start of a comment. Thus the regex ends up being ^ alone followed by a comment whose text is "!/usr/bin/perl". The net effect is that $1 evaluates to the empty string. The regular expression matches the position at the start of the string and so inserts $insertion at the start of the string. /usr/bin/perl is never matched.

Now for how to avoid such problems in the future:

• use strictures. That is, start your script with use strict; use warnings;. Had you done so you would have noticed a problem right away. In addition to the bareword in use English... (which should be use English qw(no_match_vars)), you would have seen the message alerting you to a problem in the regular expression: Unmatched ( in regex; marked by <-- HERE in m/( <-- HERE ^#!/usr/bin/perl[^\n]*\n)/.
• use \\ in double quoted strings when you want to insert an actual backslash rather than just escape the next character.
• use test strings with data before and after the insertion point, e.g. "#!/usr/bin/perl   \n\n\nyaba-dubba-doo\n";. Had you done that it would have been more obvious that it wasn't "reversing the order" but rather it was matching the start of the string alone.

And a few comments on best practice:

• avoid .*? and .* They rarely do what you want. A better choice is something more specialized like [^\n]*\n. This specifically allows one to skip past all characters on the same line and stops at the end of the line.
• avoid using '?' as a separator in substitutions (the s/// operator). At best it is confusing. '?' has a special meaning within regular expressions. Also ?pattern? "once only" searches are "vaguely deprecated". In the future you might want to consider something like s||| or s{}{}. The later can be used even with variables expressed as ${varname}. Perl knows how to handle nesting properly. See perlop for details.
• Avoid slurping except in one liners. Get in the habit of using while ($line=<INFILE>) {...} instead. It uses far less memory.
• When printing out variables use non-whitespace to surround the variable value so that you can easily see unexpected hidden whitespace. print "content=<$content>\n" is far more likely to alert you to problems that print "content=\n$content\n"

The following code illustrates these points and outputs the insertion code in the correct position:

Best, beth

Updated: explain specific cause of error and added code sample.

'better late than never' In case I have not already done so, I want to thank you for your clear and helpful feedback. I am currently experiencing some very sharp learning curves in Perl and I am only gradually being able to have the time to learn how best to utilize the Perl Monastery Gates.