regular expression in perl?

[semna]

Hi,
I have a list contains:
ID : 741158
PARENT ID : 9605
RANK : species
GC ID : 1
MGC ID : 2
SCIENTIFIC NAME : Homo sp. Altai
GENBANK COMMON NAME : Denisova hominin
//
In some part of this file the rank is subspecies. I just need those ID line (first line) which belong to species not subspecies:
I used this : my @new_list = grep {$_ =~ /\bspecies$/ && /\nID +\: *([\d]+)/} @list;
but is not working. Any suggestion? Thanks

[steven]

cat file | awk '$1=="ID"{id=$0} $1=="RANK" && $3=="species"{print id}'

Assuming an ID line is provided for each entry.
If you just need the ID number, use "id=$3" instead of "id=$0".

[semna]

Thanks Steven. But I am not sure that it works in perl .

[gringer]

So, you asked about perl. Here's a one-liner that roughly matches steven's awk script:

$ perl -ne 'if(/ID *: *(\d+)/){$id=$1};if(/RANK *: *species/){print "$id\n"}' file.txt
9605

If you're using it inside some other perl code, you'd probably do something like this:

my @new_list = ();
my $id = "";
# this assumes you want to store IDs from files or standard in
while(<>){
  if(/^ID *: *(\d+)/){$id=$1};
  if($id && /^RANK *: *species/){
    push(@new_list, $id);
  }
}

Or if you're reading from the array @list, which contains one element per line:

my @new_list = ();
my $id = "";
foreach(@list){
  if(/^ID *: *(\d+)/){$id=$1};
  if($id && /^RANK *: *species/){
    push(@new_list, $id);
  }

Or, assuming your example is exactly as your code looks (i.e. each element in @list contains a number of lines, but only one record per list element):

my @new_list = ();
foreach (@list){
  if(/^RANK *: *species/){
    if(/^ID *: *(\d+)/){
      push(@new_list, $1);
    }
  }
}

[but if that were the case, your code would probably work, but would spit out the entire record, rather than just the ID]


I would advise you to stay away from using grep in situations like this where you're modifying things inside a loop. It would do weird things like not adding to your result array if $id were 0, and changing $_ would alter your original list. See here for more information:

http://perldoc.perl.org/functions/grep.html

[steven]

Yes, that is not perl indeed but awk. If you really need to include this in a bigger perl code, the equivalent can be obtained by splitting the lines and adding a couple of "if" and "else". And I bet the same can be obtained by just using linux command lines, directly from the shell.

[steven]

Nice job gringer, that is a complete answer

[gringer]

I bet the same can be obtained by just using linux command lines, directly from the shell

I think an awk one-liner is close enough to 'linux commands, directly from the shell'. A command-line equivalent using multiple grep pipes would go something like this:

$ grep '^\(ID\|RANK\)' file.txt | grep -B 1 '^RANK : species' | grep -v '^RANK'
ID : 741158

Although you'd run into problems with that if any IDs didn't have associated ranks following them.

[semna]

Thanks so much Gringer and Steven.