FeistyLemur has asked for the wisdom of the Perl Monks concerning the following question:

This is my first post here, I have a question about something I can't explain.

I'm pretty sure I know what I did wrong, I'm just not 100% sure why it was wrong and wanted to clarify to better my understanding of regex in perl, as I'm fairly inexperienced and looking to improve.

I was searching for vlan ids for removal in the output of "ip addr" in the following way.

my $ethernet=`ip addr`;
if ($ethernet!~/eth1\.\d{4}\@/gm){
    print "No vlans exist on device.\n";
    exit;
}
while ($ethernet=~/(eth1\.\d{4})\@/gm) {
    print `ip link del $1\n";
    print "Removed $1\n";
}

[download]

Doing this as above mostly worked, but it would always miss the first entry for no reason I can explain. So if there were 25 vlan entries in the string from 1001 to 1025 it would match and delete 1002-1025 without fail, and miss 1001 every time.

Changing line 2 to:

if ($ethernet!~/eth1\.\d{4}\@/){

Does what I intended to do.

I had the same problem with a similar line I was using to scrub IP addresses off the vlan devices again missing the first match, I just don't understand why the while loop breaks in the way it does because of the /gm flag on the preceeding if statements match, and was hoping someone could explain. Thanks in advance.

Replies are listed 'Best First'.
Re: A regex question.
by AnomalousMonk (Archbishop) on Jul 31, 2015 at 22:00 UTC

    Because you're using the  /g regex modifier for all matches, the string match position never resets to the start of the string, but advances for each and every match performed. (Update: I should have written: advances for each and every successful match performed.) 

    c:\@Work\Perl\monks>perl -wMstrict -le
"my $s = 'xx111yy22xx333yy44zz555xx666';
 ;;
 print qq{first capture: '$1'} if $s =~ m{ (\d+) }xmsg;
 ;;
 print qq{subsequent captures: '$1'} while $s =~ m{ (\d+) }xmsg;
"
first capture: '111'
subsequent captures: '22'
subsequent captures: '333'
subsequent captures: '44'
subsequent captures: '555'
subsequent captures: '666'

    [download]
    The fact that the first match in the OPed code is a negative | negated match (so no action is taken) does not matter; it's still a  /g match against the same string. Eliminating the  /g modifier on the first match will cause string match position to be reset before subsequent matches (which must still use /g).

    Note that use of the  /g modifier in an
        if ($string =~ m{ ... }xmsg) { ... }
    statement is useless except as a way to introduce exactly the behavior you are seeing!

    Update: A number of incremental updates made.

    Give a man a fish:  <%-(-(-(-<

[reply]
[d/l]
[select]

      Hm, well I really didn't understand that about how the match operator works so I guess that would explain it. I had assumed /g was pointless on the if after thinking about it, but the part I really couldn't figure out was why the negative match would affect the subsequent one in any way.

[reply]

        Hello FeistyLemur, and welcome to the Monastery!

        See the section “Global matching” in perlretut#Using-regular-expressions-in-Perl, and in particular note the following (emphasis added.):

        In scalar context, successive invocations against a string will have //g jump from match to match, keeping track of position in the string as it goes along....
        A failed match or changing the target string resets the position.... The current position in the string is associated with the string, not the regexp.

        Hope that helps,

        Athanasius <°(((>< contra mundum Iustus alius egestas vitae, eros Piratica,

[reply]
[d/l]
        ... the negative match ...

        It's important to understand that the negation is made on the result of the  /eth1\.\d{4}\@/gm match against the  $ethernet string. The match itself is actually successful! It's as if the
            if ($ethernet !~ /eth1\.\d{4}\@/gm) { ... }
        statement had been written as
            if ( ! ($ethernet =~ /eth1\.\d{4}\@/gm)) { ... }
        (note =~ vice !~). In fact, that's pretty much the way Perl sees the code: 

        c:\@Work\Perl\monks>perl -wMstrict -MO=Deparse,-p -le
"my $ethernet=`ip addr`;
 if ($ethernet!~/eth1\.\d{4}\@/gm){
     print qq{No vlans exist on device.\n};
     exit;
 }
"
BEGIN { $^W = 1; }
BEGIN { $/ = "\n"; $\ = "\n"; }
use strict 'refs';
(my $ethernet = `ip addr`);
unless (($ethernet =~ /eth1\.\d{4}\@/gm)) {
    print("No vlans exist on device.\n");
    exit;
}
-e syntax OK

        [download]

        I had assumed /g was pointless on the if ...

        Given what you were trying to do, it was pointless, but it was not without effect!

        Give a man a fish:  <%-(-(-(-<

[reply]
[d/l]
[select]
Re: A regex question.
by Aldebaran (Curate) on Aug 01, 2015 at 08:52 UTC

    It seems to me that your problem is in your control mechanisms, specifically that your if statement should be inside of a while statement that goes through your data line by line. Also maybe a global variable to keep track of matches. If you're reading a file, then you'll have something like:

    my $matches = 0;
open FILE, ">filename.txt" or die $!;
while (my $ethernet = <FILE>) {

    [download]
    if ($ethernet=~/eth1\.\d{4}\@/m){
  ## do something
  matches++;
}

}

    [download]
[reply]
[d/l]
[select]

      This won't work because you are opening the filehandle only for writing. In fact it will clobber the contents of the file.

      It should be: 

      # open FILE, ">filename.txt" or die $!; #output only, and deprecated s
+yntax

open my $file, '<', 'filename.txt' or die $!;
while (my $ethernet = <$file>) { 

  ## do something

}

      [download]
      The way forward always starts with a minimal test.
[reply]
[d/l]

Back to Seekers of Perl Wisdom