bfilter/trunk/bfilter.pl

#!/usr/bin/perl -w
#

use strict;
use locale;

# maximum entries
my $max = 0;
# minimum letters to search by
my $min_len = 2;

sub print_file {
        my $f = shift || return;
        open(F, $f) || die "$f: $!";
        while(<F>) {
                print;
        }
        close(F);
}

print qq{
var headlines = Array();
};

my @part_arr;
my $last_part = '';
my $total = 0;

while(<STDIN>) {
        chomp;

        # escape single quote
        s/'/\\'/g;

        # split into min_len part and rest
        my ($part,$rest) = ( substr($_,0,$min_len), substr($_,$min_len) );

        # make part lowercase
        $part = lc($part);

        $last_part = $part if (! $last_part);

        # new part?
        if ($part ne $last_part) {
                print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
                $total += $#part_arr;
                @part_arr = ();
                $last_part = $part;
        }
        push @part_arr, "'$_'";

        # break out?
        last if ($max && $total > $max);
}

print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
print "var min_len = $min_len;\n";
print "// index elements: $total\n";
1	dpavlin	1	#!/usr/bin/perl -w
2			#
3
4			use strict;
5			use locale;
6
7			# maximum entries
8			my $max = 0;
9			# minimum letters to search by
10			my $min_len = 2;
11
12			sub print_file {
13			my $f = shift \|\| return;
14			open(F, $f) \|\| die "$f: $!";
15			while(<F>) {
16			print;
17			}
18			close(F);
19			}
20
21			print qq{
22			var headlines = Array();
23			};
24
25			my @part_arr;
26			my $last_part = '';
27			my $total = 0;
28
29			while(<STDIN>) {
30			chomp;
31
32			# escape single quote
33			s/'/\\'/g;
34
35			# split into min_len part and rest
36			my ($part,$rest) = ( substr($_,0,$min_len), substr($_,$min_len) );
37
38			# make part lowercase
39			$part = lc($part);
40
41			$last_part = $part if (! $last_part);
42
43			# new part?
44			if ($part ne $last_part) {
45			print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
46			$total += $#part_arr;
47			@part_arr = ();
48			$last_part = $part;
49			}
50			push @part_arr, "'$_'";
51
52			# break out?
53			last if ($max && $total > $max);
54			}
55
56			print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
57			print "var min_len = $min_len;\n";
58			print "// index elements: $total\n";