in reply to Re: Problems counting regex matches
in thread Problems counting regex matches

... this won't work for all cases though, if you have two matching words in the neighbourhood of the same (currency|foreign exchange), just one will be counted. For exemple in "Currency revenue sales growth", you'll just get "revenue" ...

The following works for overlapping matches. It also needs 5.10+ because in addition to (?|pattern), it uses (*FAIL) from the Special Backtracking Control Verbs introduced in that version. The variation that only counts occurrences may be a little faster. 

use 5.010;  # for (?|pattern) and (*FAIL)

use warnings
    FATAL => 'all'
    ;
use strict;


# factored regexes

my $r_s_g = qr{ (?i) (?> \b (?: revenues? | sales | growth    ) \b) }x
+ms;
my $c_fe  = qr{ (?i) (?> \b (?: currency | foreign \W exchange) \b) }x
+ms;

my $word = qr{ \b \w+ \W+ (?: \b | \Z) }xms;

my $max_between = 4;
my $near = qr{ (?:$word){0,$max_between}? }xms;


# the test text

my $s = <<EOT;
Currency
revenue and sales growth in
foreign exchange Sales and
Revenues
EOT
print qq{[[$s]] \n\n};


# extract matches 'in context'

our @matches;
use re 'eval';
$s =~ m{
    (?=
      (?| ($r_s_g) \W+ $near ($c_fe) | ($c_fe) \W+ $near ($r_s_g))
      (?{ push @matches, [ $1, $2 ] })
      (*FAIL)
    )
  }xmsg;

print qq{'$_->[0]' ... '$_->[1]' \n} for @matches;
print qq{\n};


# just count matches

our $n_matches;
$s =~ m{
    (?=
      (?| $r_s_g \W+ $near $c_fe | $c_fe \W+ $near $r_s_g)
      (?{ ++$n_matches })
      (*FAIL)
    )
  }xmsg;

print qq{$n_matches matches \n};

[download]

Output: 

[[Currency
revenue and sales growth in
foreign exchange Sales and
Revenues
]]

'Currency' ... 'revenue'
'Currency' ... 'sales'
'Currency' ... 'growth'
'revenue' ... 'foreign exchange'
'sales' ... 'foreign exchange'
'growth' ... 'foreign exchange'
'foreign exchange' ... 'Sales'
'foreign exchange' ... 'Revenues'

8 matches

[download]

In Section Seekers of Perl Wisdom