Multiline regex for moving lines?

bowei_99 has asked for the wisdom of the Perl Monks concerning the following question:

I'm trying to figure out a multiline regex to:

1. Parse the following text:

#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
[download]

2. Move the third set of lines, i.e.:

#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
[download]

to the top, right under #EXTM3U, so that it would show up as line 2. The other lines should be kept the same, just shifted down.

I've written the following:

$text = <<'LIST';
#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
LIST

$text =~ m{ (\#EXT-X-STREAM-INF:PROGRAM-ID=
                \d+,
                BANDWIDTH=\d+ \s*
                \d+\.m3u8)
         }xms;

print "1 - $1, 2 - $2, 3 - $3\n";
[download]

which yields:

# ./test.pl
1 - #EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8, 2 - , 3 -
[download]

However, I'm not sure how to specify in the regex to split by: a) the first chunk being the top two entries, b) the second chunk being what I want to move, c) the last chunk. I figure once I have that, it's a simple matter of specifying $1, $2 and $3 in the replacement part of the regex.

Note that I realize I could split this into an array, but I'm trying to find an efficient regex to do the job here. Thoughts?

-- Burvil

Comment on Multiline regex for moving lines? Select or Download Code

Replies are listed 'Best First'.

Re: Multiline regex for moving lines?
by kcott (Archbishop) on Sep 05, 2013 at 02:42 UTC

G'day bowei_99,

Your text is made up of a series of elements each consisting of a hash (#) followed by a variable number of characters which aren't hashes, i.e. "[#][^#]+".

The "last chunk", as you called it, doesn't need to be part of the replacement. Just match the first "#...", a pair of "#...", the "#..." you want to move; then move the 3rd capture before the 2nd:

$ perl -Mstrict -Mwarnings -le '
    my $text = <<LIST;
#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
LIST
    $text =~ s{\A([#][^#]+)((?:[#][^#]+){2})([#][^#]+)}{$1$3$2}m;
    print $text;
'
#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
[download]

-- Ken

[reply]
[d/l]
[select]

Re: Multiline regex for moving lines?
by Athanasius (Archbishop) on Sep 05, 2013 at 03:01 UTC

For flexibility, here’s an approach which accepts the number of the entry to be moved up as a command line parameter:

#! perl
use strict;
use warnings;

my $target = $ARGV[0] - 1;
my $text   = <<'LIST';
#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
LIST

my $entry = qr{ \#EXT-X-STREAM-INF:PROGRAM-ID= \d+ 
                ,BANDWIDTH= \d+ \s+ \d+\.m3u8 \s+ }xms;

if ($text =~ m{
                ( \A  \#EXTM3U \s+     )    #1
                ( (?: $entry){$target} )    #2
                ( (?: $entry)          )    #3
                ( (?: $entry)*         )    #4
              }xms)
{
    print "$1$3$2$4";
}
else
{
    print "No match found\n";
}
[download]

Output:

12:57 >perl 710_SoPW.pl 3
#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8

12:57 >
[download]

Hope that helps,

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

[reply]
[d/l]
[select]

Re: Multiline regex for moving lines?
by johngg (Canon) on Sep 05, 2013 at 08:15 UTC

Rather than using a complicated regex you could read your records into an array then print in the desired order using an array slice.

use strict;
use warnings;

open my $inFH, q{<}, \ <<EOF or die $!;
#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
EOF

my @records;

push @records, scalar <$inFH>;
push @records, join q{}, map scalar <$inFH>, 1 .. 2
   while not eof $inFH;

print for @records[ 0, 3, 1, 2, 4, 5 ];
[download]

The output.

#EXTM3U
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=700000
700.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=80000
80.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=400000
400.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1500000
1500.m3u8
#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2500000
2500.m3u8
[download]

I hope this is helpful.

Cheers,

JohnGG

[reply]
[d/l]
[select]

Re: Multiline regex for moving lines?
by Anonymous Monk on Sep 05, 2013 at 02:33 UTC

Note that I realize I could split this into an array, but I'm trying to find an efficient regex to do the job here. Thoughts?

How would you do it using split into an array?

The logic isn't too different using regex (maybe), just the language is different(regex, 1,2 )

I imagine something along the lines of (you fill in the blanks )

my $lineregex = qr{}ms;
$input =~ s{
( # $1 is first 3 lines
  $line   $line   $line
)
( # $2 is second 3 lines
  $line   $line   $line
)
( # $3 is third 3 lines
  $line   $line   $line
)

}{$2$3$1}gx;
[download]

[reply]
[d/l]

Re: Multiline regex for moving lines?
by Laurent_R (Canon) on Sep 05, 2013 at 06:13 UTC

You can do it with regexes, and have been shown several ways of doing it, but regexes is not the best tool to do that. Just as you can drive a nail into a piece of wood by hammering it with the handle of a screwdriver.

Splitting your input into an array of lines seems more natural, more flexible if your rules change and would scale better if your input grows larger. So why not using split?

[reply]