Matching lines by number of words with regex

Question

I am still new to regex and I've run into a bit of a problem. I am building a parsing script and I need to to be able to pull out lines with a certain length out of a file.

How would I write a regex to match lines that have a certain number of words? Eg I want to match all lines in a file that have 3 words.

Could I extend that to find all lines within certain parameters? Eg I want to match all lines in a file that have between 2 and 5 words.

I am using perl in case that matters. Thanks!

Answer 1

This depends on what you consider to be a word. Perl 5 considers a word to be /\w+/. If you have a different definition you will need to supply it.

You can find the number of times a regex matched by using the Count Of secret operator: ()=:

my $count = ()= $line =~ /\w+/g;

Once you know the number of words, you can easily construct an if statement to print a line if the number or words is between two numbers using the >= and <= operators.

In Perl 5.10 and later, it is possible to match two to five words using the possessive quantifier:

#!/usr/bin/perl

use strict;
use warnings;

while (my $line = <DATA>) {
    next unless $line =~ /^(?:\W*+\w++){2,5}$/;
    print $line;
}

__DATA__
one
one two
one two three
one two three four
one two three four five
one two three four five six

Answer 2

(Chas's answer wasn't quite right -- he missed a flag on the m// operator.) :)

use strict;
use warnings;

use Data::Dumper;

my @good;
foreach my $line (<DATA>)
{
    chomp $line;
    my $matches =()= ($line =~ /\b\w+\b/g);
    print "(debugging) found matches $matches\n";
    push @good, $line if $matches >= 2 and $matches <= 5;
}

print "matching lines: ", Dumper(\@good);

__DATA__
foo bar baz bap
foo bar baz
blah blah blah foooo

bip

produces:

(debugging) found matches 4
(debugging) found matches 3
(debugging) found matches 4
(debugging) found matches 0
(debugging) found matches 1
matching lines: $VAR1 = [
          '    foo bar baz bap',
          '    foo bar baz',
          '    blah blah blah foooo'
        ];

Answer 3

Replace the 3 with how many words you are looking for. This regex assumes no spaces or tabs start the line:

^(?=(\b[A-Za-z0-9.]+\b[\x20]){3})(.)*

This says match: from the beginning of each line look through each line for 3 alpha numeric or period words are each trailed by a single space and if what we looked ahead for matches then select the entire line no matter what is on it

Note: the \x20 matches a space character and the regex was developed in notepad++ by memory and hand.

Answer 4

Here's a KISS way.

while(<>){
  #assumption: words separated by spaces
  @s = split /\s+/ ;
  # now check the length of @s and do if/else
}