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 = 3;

my $debug = 1;

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 STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug);
                print "headlines['$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	dpavlin	3	my $min_len = 3;
11	dpavlin	1
12	dpavlin	3	my $debug = 1;
13
14	dpavlin	1	sub print_file {
15			my $f = shift \|\| return;
16			open(F, $f) \|\| die "$f: $!";
17			while(<F>) {
18			print;
19			}
20			close(F);
21			}
22
23			print qq{
24			var headlines = Array();
25			};
26
27			my @part_arr;
28			my $last_part = '';
29			my $total = 0;
30
31			while(<STDIN>) {
32			chomp;
33
34			# escape single quote
35			s/'/\\'/g;
36
37			# split into min_len part and rest
38			my ($part,$rest) = ( substr($_,0,$min_len), substr($_,$min_len) );
39
40			# make part lowercase
41			$part = lc($part);
42
43			$last_part = $part if (! $last_part);
44
45			# new part?
46			if ($part ne $last_part) {
47	dpavlin	3	print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug);
48	dpavlin	1	print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
49			$total += $#part_arr;
50			@part_arr = ();
51			$last_part = $part;
52			}
53			push @part_arr, "'$_'";
54
55			# break out?
56			last if ($max && $total > $max);
57			}
58
59			print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
60			print "var min_len = $min_len;\n";
61			print "// index elements: $total\n";