This should work:

    use HTML::Parser;

    use Symbol;
    my $TXT = gensym;

    my $parser = HTML::Parser->new(api_version => 3,
        text_h => [ sub {
            print $TXT @_;
        }, "dtext" ]);

    for my $html (@ARGV) {
        (my $txt = $html) =~ s/\.html$/\.txt/;
        open $TXT, ">$txt" or die "Can't open $txt: $!";
        $parser->parse_file($html);
        close $TXT or die "Can't close $txt: $!";
    }
[download]

Yes, I know--I changed a bunch of stuff around. But I find it easier to work w/ the new HTML::Parser API rather than the old. In the new API you set up handlers for specific events, rather than subclassing and adding your own functionality.

So we create a new parser object and register an event to handle text (just like your text method) by using "text_h" (text handler). We give it a subref to run on that event, then a list of arguments to pass to our subref. "dtext" means text that's been passed through decode_entities, so we don't have to do that ourselves anymore.

That subref is a closure referring to the $TXT handle, which we open and close each time through the loop. There may be a more elegant way to do this, but it seemed to work for me.

I wanted to use the -i CL option, but you can't specify the name of the *new* file, only the name of the backup file. So we're stuck with looping through @ARGV and messing about manually with each of the files; but that's not so bad. Unless, of course, I've messed it up. :)

Each time through the loop we grab the name of the file, then change the .html extension to .txt.

Save this script and run it like this:

    % ./foo.pl file1.html file2.html ...
[download]