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

Greetings fellow monks,

A while back, I sought wisdom regarding a RegEx filter that would filter \s+ from a scalar but skip filtering between to marker labels. I got several really great replies, but I've wandered into a new problem: I need the same filter but allow for multiple marker labels. Here are the details:

I have an large string scalar and I'd like to filter out multiple white spaces, converting them into just a single space per instance. Here's an example of the scalar:

A Bridge Too Far     Hosted by Rod Stuart     Friendly Skys
42    STARTPRESERVE1 Life, the universe...   and Everything STOPPRESER
+VE1   Bob
HotWheels are cool     More movies on Fox           File server
A Bridge Too Far     Hosted by Rod Stuart     Friendly Skys
42    STARTPRESERVE2 Life, the universe...   and Everything STOPPRESER
+VE2   Bob
HotWheels are cool     More movies on Fox           File server

[download]

Sounds like a job for $text =~ s/\s+/ /g;, except that I want to avoid the filtering between multiple START and STOP markers. Here's the original code that uses a single pair of markers (provided by IO):

$text =~ s/\s+|(STARTPRESERVE.*?STOPPRESERVE)/${[$1,' ']}[!$1]/gs;

This works great exept that I now need multiple markers like STARTPRESERVE1 .. 3 and so forth. My first thought was a loop through an array of these markers doing a regex, but of course that won't work. So I think it needs to be a single regex. (Right?) So here's my feeble attempt: (Please try not to laugh.)

$$text_ref =~ s%
  (STARTPRESERVE1.*?STOPPRESERVE1)
|
  (STARTPRESERVE2.*?STOPPRESERVE2)
|
  (PRESERVESTART.*?PRESERVESTOP)
|
  \s+
%
  ((defined $1 and ($1 =~ /^\s+$/)) ? ' ' : $1) .
  ((defined $2 and ($2 =~ /^\s+$/)) ? ' ' : $2) .
  ((defined $3 and ($3 =~ /^\s+$/)) ? ' ' : $3)
%eigsx;

[download]

Am I headed in the right direction, or am I missing an easier solution? Thanks all.

gryphon
code('Perl') || die;

Replies are listed 'Best First'.
Re: RegEx filter \s ! between labels, part 2
by fglock (Vicar) on Oct 02, 2002 at 19:55 UTC

    Previous code was eliminating newlines. This one is matching a literal "space" instead of "\s".

    It is better documented, too :)

    $text =~ s/
        [ ]+ |                  # spaces, or:
        (                       # begin $1
            STARTPRESERVE(\d?)  #     STARTPRESERVE + $2
            .*?                 #     anything
            STOPPRESERVE\2      #     STOPPRESERVE + $2
        )                       # end $1
    /${ [$1,' '] }[ ! $1 ]/sgx;

# [$1,' '] is an unnamed array.
# ! $1 is:
#     0 if $1 exists;
#     1 if spaces were found.

    [download]
[reply]
[d/l]

      Er... OK, but this looks and functions like what I had before. How do I incorporate multiple PRESERVE tags? For example, I need to s/\s+/ /g everywhere except between MARKA and MARKB, between STARTPRESERVE and STOPPRESERVE, and between YETANOTHERMARKER and YETANOTHERMARKEREND.

      I can't loop through an array of regexes because they'd just cancel eachother out and do the s/\s+/ /g everywhere. I think I need a single regex, or perhaps a pulling appart of the string into an array split by the various markers, then each element analysed. What do you think is the best approach?

      ## START PSEUDO-CODE
my $string = $big_string_from_my_original_example;
my @array_of_stuff = split
  /\b(?:STARTPRESERVE)|(?:STOPPRESERVE)\b|
  \b(?:MARKA)|(?:MARKB)\b/, $string);
foreach (@array_of_stuff) {
  if (&check_for_marker) {
    push @new_array, $_;
    next;
  }
  s/\s+/ /g;
  push @new_array, $_;
}
$string = join '', @new_array;
## END PSEUDO-CODE

      [download]

      Oh, and I actually do need to remove line breaks but keep the markers in place. And although it's theoretically possible, I highly doubt any markers will ever be nested. (If they are, it's due to user error.)

      gryphon
      code('Perl') || die;

[reply]
[d/l]
Re: RegEx filter \s ! between labels, part 2
by fglock (Vicar) on Oct 02, 2002 at 18:00 UTC

    This is a bit weird, but it seems to work:

    $text =~ s/\s+|(STARTPRESERVE(\d?).*?STOPPRESERVE$1)/${[$1,' ']}[!$1]/gs;

    update: see merlin below. Now it works with nested PRESERVEs:

    $text =~ s/\s+|(STARTPRESERVE(\d?).*?STOPPRESERVE\2)/${[$1,' ']}[!$1]/
+gs;

    [download]
[reply]
[d/l]
[select]
[reply]
[d/l]
[select]
Re: RegEx filter \s ! between labels, part 2
by fglock (Vicar) on Oct 03, 2002 at 18:07 UTC

    Ok, here it goes with these specifications:

    %Marks = (
  STARTPRESERVE => 'STOPPRESERVE',
  BEGIN => 'END' );
$MarkOptions = join("|",keys %Marks);

$text =~ s/
        \s+ |                   # spaces, or:
        (                       # begin $1
            ($MarkOptions)(\d?) #     $2=START + $3=optional_number
            .*?                 #     anything
            $Mark{\2}\3         #     $Mark{START} + optional_number
        )                       # end $1
    /${ [$1,' '] }[ ! $1 ]/sgx;

# [$1,' '] is an unnamed array.
# ! $1 is:
#     0 if $1 exists;
#     1 if spaces were found.

    [download]

    Note: you may add \b to the markers if you feel that you need it.
    I kept the optional number after the marker, you may remove it if you don't need it.

[reply]
[d/l]

Back to Seekers of Perl Wisdom