How to trigger Perl multiline substitution

Question 1

What you need is the -0777 switch which will cause the entire file to be read into a single string. If this is not used, the files will be read in line-by-line mode, and you can never match a multi-line statement that way.

Also, as Andomar points out, you are missing the -p switch, but I assume you figured that out.

The modifiers on the regex won't matter in this case, except the /g modifier. /m only affects ^ and $, and /s causes wildcard . to also match newlines. None of this applies to your regex.

So basically, you want something like:

perl -0777 -pi -e 's/<![^>]+>//g' ...

Side note:

Html should be handled with parsers, ideally, so I spent a few minutes working on using HTML::Parser which has a convenient option to strip declarations by adding a handler. Something like this seems to print ok for a single file:

perl -MHTML::Parser -we '
    $p = HTML::Parser->new(default_h => [sub {print @_},'text'] ); 
    $p->handler(declaration => ''); 
    $p->parse_file(shift) or die $!; " yourfile.html

I figured it would be overkill so I abandoned trying to fix it with the -pi in-place edit switches, but it is (probably) easily implemented in a script.

Question 2

First, you seem to be missing the -p parameter, for processing the input line by line. -i doesn't seem to do much without -p.

Second, since -pi processes the input line-by-line, it can't replace a regex that spans more than one line.

You could write a Perl script instead. This script should run your regex on the entire content of all files passed on the command line:

use IO::All;

foreach my $file (@ARGV) {
    my $content = io($file)->slurp;
    $content =~ s/<![^>]+>//g;
    $content > io($file);
}

The command cpan IO:All should install the IO:All module, if it is not present on your system.

P.S. The m and s options only affect ., ^ and $. I think you can omit them.