one liner to extract data block by block

Question

I always deal with data files that consist of many data blocks of the following format:

*name* attr (
        VALID (
                late_lead_up xxx ar uclk reff xxx slope xxx
                late_lead_dn xxx af uclk reff xxx slope xxx
                early_trail_up xxx af uclk reff xxx slope xxx
                early_trail_dn xxx ar uclk reff xxx slope xxx
              )
        CEXT xxx
        CREF xxx
        REFF xxx
        QUALIFIED_CLOCK
)

Is there anyway I can extract the "name" that I interested in using a one-liner from the command line?

Answer 1

If your block always start with '*name* attr (' and always ends with a ')' alone with no leading space, you can try (given that foo is the block name and data.txt is the file to parse):

awk '/ attr \($/ {if($1==n)b=1}  {if(b)print}  /^\)$/ {b=0}' n=foo data.txt

Answer 2

Well, you tagged it Perl, so here is how I would do it in Perl:

#!/usr/bin/perl

use strict;
use warnings;

die "usage: $0 name datafile\n    or cat datafile | $0 name\n" 
    unless @ARGV > 0;

my $name = shift;
my $re   = qr/\A$name attr/; 

my $rec = '';
while (my $line = <>) {
    $rec .= $line;
    next unless $line =~ /^\)/;
    print $rec if $rec =~ /$re/;
    $rec = '';
}

You could turn it into a one-liner like this

perl -ne '$a.=$_;next unless/^\)/;print$a if$a=~/^name/;$a=""' datafile

but I prefer the script. Remember to replace name with the name of the record.

Answer 3

Using this file for demo purposes:

of_interest attr (
    1:VALID (
        1:late_lead_up xxx ar uclk reff xxx slope xxx
        1:late_lead_dn xxx af uclk reff xxx slope xxx
        1:early_trail_up xxx af uclk reff xxx slope xxx
        1:early_trail_dn xxx ar uclk reff xxx slope xxx
    1:)
    1:CEXT xxx
    1:CREF xxx
    1:REFF xxx
    1:QUALIFIED_CLOCK
)

boring attr (
    2:VALID (
        2:late_lead_up xxx ar uclk reff xxx slope xxx
        2:late_lead_dn xxx af uclk reff xxx slope xxx
        2:early_trail_up xxx af uclk reff xxx slope xxx
        2:early_trail_dn xxx ar uclk reff xxx slope xxx
    2:)
    2:CEXT xxx
    2:CREF xxx
    2:REFF xxx
    2:QUALIFIED_CLOCK
)

of_interest attr (
    3:VALID (
        3:late_lead_up xxx ar uclk reff xxx slope xxx
        3:late_lead_dn xxx af uclk reff xxx slope xxx
        3:early_trail_up xxx af uclk reff xxx slope xxx
        3:early_trail_dn xxx ar uclk reff xxx slope xxx
    3:)
    3:CEXT xxx
    3:CREF xxx
    3:REFF xxx
    3:QUALIFIED_CLOCK
)

This one-liner (split for readability):

awk '
    BEGIN               {s=0}
    /^of_interest /     {s=1}
    /^)$/               {if (s==1) {print};s=0}
                        {if (s==1) print}'

or the minimum character version:

awk 'BEGIN{s=0}/^of_interest /{s=1}/^)$/{if(s==1){print};s=0}{if(s==1)print}'

gives you:

of_interest attr (
    1:VALID (
        1:late_lead_up xxx ar uclk reff xxx slope xxx
        1:late_lead_dn xxx af uclk reff xxx slope xxx
        1:early_trail_up xxx af uclk reff xxx slope xxx
        1:early_trail_dn xxx ar uclk reff xxx slope xxx
    1:)
    1:CEXT xxx
    1:CREF xxx
    1:REFF xxx
    1:QUALIFIED_CLOCK
)
of_interest attr (
    3:VALID (
        3:late_lead_up xxx ar uclk reff xxx slope xxx
        3:late_lead_dn xxx af uclk reff xxx slope xxx
        3:early_trail_up xxx af uclk reff xxx slope xxx
        3:early_trail_dn xxx ar uclk reff xxx slope xxx
    3:)
    3:CEXT xxx
    3:CREF xxx
    3:REFF xxx
    3:QUALIFIED_CLOCK
)

which I believe is what you were after.

It's basically a simple state machine that turns on printing when it finds the desired block start and turns it off when it finds the end of that block.

UPDATE: Here's a perl one-liner that takes care of your qualified_clock requirement. Enjoy :-)

perl -e '$s=1;while(<STDIN>){if(/^of_interest /){$s=1;$f=0;$x="";}if(($s==1)&&/QUALIFIED_CLOCK/){$f=1;}if(/^\)$/){if($s==1){$x.=$_;}if($f==1){print$x;}$s=0;next;}if($s==1){$x.=$_;}}'

Answer 4

Here is one way to to it as a Perl one-liner:

perl -ne '$m = 1 if /^insert_name_here attr/; print if $m; $m = 0 if /^\)$/' file.txt

Answer 5

I see in your comments to another answer that you also want to search within the block for a string like 'QUALIFIED_CLOCK'.

In that case, if your data blocks are separated by a blank line you can use Perl's paragraph mode to read it in blocks and print out the ones you are interested in. For example:

perl -00 -ne 'print if /^block_name/ and /QUALIFIED_CLOCK/' file.txt

This is also possible in awk by setting RS.

Answer 6

Far less characters and simpler solution than Pax's one

perl -ne '/^of_interest /../^\)/ and print'

or

awk '/^of_interest /,/^\)/{print}'

or

sed -n '/^of_interest /,/^)/p'