Question

Most examples of inplace editing are one-liners that iterate through a file or files, reading and printing one line at a time.

I can’t find any examples of reading an entire file into an array, modifying the array as needed, and then printing the array while using the ^I switch to do an inplace edit. When I try to read the entire file from the diamond operator, edit the contents and print the entire contents, I find that the print goes to STDOUT instead of ARGVOUT and that ARGVOUT is closed. I can open the same file for output and then print to it, but I’m not sure I understand why that is necessary. Here is an example:

#!/usr/bin/perl
use strict;
use warnings;
use 5.010;

my $filename = 'test.txt';

push @ARGV, $filename;

$^I = ".bk";

my @file = <>; #Read all records into array
chomp @file;
push @file, qw(add a few more lines);

print join "\n", @file; #This prints to STDOUT, and ARGVOUT is closed. Why?

Running the above makes a backup of the test.txt file as expected, but leaves the edited test.txt empty, printing the edited contents to STDOUT instead.

Answer 1

See perlrun.

When the -i switch has been invoked, perl starts the program using ARGVOUT as the default file handle instead of STDOUT. If there are multiple input files, then every time the <> or <ARGV> or readline(ARGV) operation finishes with one of the input files, it closes ARGVOUT and reopens it to write to the next output file name.

Once all the input from <> is exhausted (when there are no more files to process), perl closes ARGVOUT and restores STDOUT as the default file handle again. Or as perlrun says

#!/usr/bin/perl -pi.orig
s/foo/bar/;

is equivalent to

#!/usr/bin/perl
$extension = '.orig';
LINE: while (<>) {
    if ($ARGV ne $oldargv) {
        if ($extension !~ /\*/) {
            $backup = $ARGV . $extension;
        }
        else {
            ($backup = $extension) =~ s/\*/$ARGV/g;
        }
        rename($ARGV, $backup);
        open(ARGVOUT, ">$ARGV");
        select(ARGVOUT);
        $oldargv = $ARGV;
    }
    s/foo/bar/;
}
continue {
    print;  # this prints to original filename
}
select(STDOUT);

Once you say my @file = <> and consume all the input, Perl closes the filehandle to the backup files and starts directing output to STDOUT again.

---

The workaround, I think, is to call <> in scalar context and check eof(ARGV) after each line. When eof(ARGV)=1, you have read the last line in that file and you get one chance to print before you call <> again:

my @file = ();
while (<>) {
    push @file, $_;
    if (eof(ARGV)) {
        # done reading current file
        @processed_file = &do_something_with(@file);
        # last chance to print before ARGVOUT gets reset
        print @processed_file;
        @file = ();
    }
}