Perl autosplit with non-file arguments - perl

Question

I'm trying to create a perl script that autosplits the STDIN, and then does something to column X. But I want the column to be passed by argument to the script. However, apparently, when I invoke perl with the autosplit flag, it switches to a mode where it tries to implicitly open the "files" defined in the command line arguments, even though my arguments aren't files.
Example:
my $column = shift;
while(<STDIN>) {
print "F $column: $F[$column]!\n";
}
Then I try to run it with argument 2 to print the 2nd column:
$ echo -e "1 22\n2 33\n3 55" | perl -a myscript.pl 2
When I turn on warnings and such, I get this error message:
Can't open 2: No such file or directory (#1)
(S inplace) The implicit opening of a file through use of the <>
filehandle, either implicitly under the -n or -p command-line
switches, or explicitly, failed for the indicated reason. Usually
this is because you don't have read permission for a file which
you named on the command line.
(F) You tried to call perl with the -e switch, but /dev/null (or
your operating system's equivalent) could not be opened.
If I omit the -a, I don't get any errors and all is good, but then I need to manually split my lines. Is there any other flag I can use to make autosplit not do that? Or alternatively, can I start autosplit inside the script, so that it won't try to implicitly open the arguments as files? I can manually split my input, I just wanted to know what other alternatives I have.

Answer 1

There's not much of a "mode" here. -n adds
while (readline) {
...
}
around your code; -a adds a split statement:
while (readline) {
our #F = split;
...
}
readline (without argument) reads from the filenames specified in #ARGV (or STDIN if #ARGV is empty).
-a implicitly enables -n because otherwise there is nothing for it to do; there is no "autosplit mode", it simply adds code around your program.
Your best bet is to split manually:
while (<STDIN>) {
my #F = split;
print "F $column: $F[$column]!\n";
}
It's fairly short and it makes it obvious what's going on.
I don't like using the command line shortcut switches (e.g. -a, -p, etc.) in script files. In a file you have enough space to write multiple lines, properly formatted / indented. It's better to be explicit.
That said, you can technically do this:
#!perl -na
our $column;
BEGIN { $column = shift; }
print "F $column: $F[$column]!\n";
This is equivalent to:
while (readline) {
our #F = split;
our $column;
BEGIN { $column = shift; }
print "F $column: $F[$column]!\n";
}
The BEGIN block only runs once, at compile time. Provided the script is only called with one argument, this will remove it from #ARGV before the readline loop starts, which will make it read from STDIN.

Answer 2

-n, -p, -l, -a and -F don't change the state of the Perl interpreter. They literally alter the source code.
$ perl -MO=Deparse -a -ne'f()'
LINE: while (defined($_ = readline ARGV)) {
our #F = split(' ', $_, 0);
f();
}
-e syntax OK
As you can see, there's really no such thing as autosplit. It's just a normal split injected into the source code, and it makes no sense to ask Perl to insert a split instead of just inserting a split. So, simply add the following to your loop:
my #F = split;