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;
# if more than x elements, warn to increase min_len
my $increase_at = 500;

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;

my $max_elements = 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);
                $max_elements = $#part_arr if ($#part_arr > $max_elements);
                print "headlines['$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";

print STDERR "You have more than $increase_at elements in one array,\nyou whould probably increase min_len to ",$min_len+1," or higher.\n" if ($max_elements > $increase_at);
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	# if more than x elements, warn to increase min_len
12	my $increase_at = 500;
13
14	my $debug = 1;
15
16	sub print_file {
17	my $f = shift \|\| return;
18	open(F, $f) \|\| die "$f: $!";
19	while(<F>) {
20	print;
21	}
22	close(F);
23	}
24
25	print qq{
26	var headlines = Array();
27	};
28
29	my @part_arr;
30	my $last_part = '';
31	my $total = 0;
32
33	my $max_elements = 0;
34
35	while(<STDIN>) {
36	chomp;
37
38	# escape single quote
39	s/'/\\'/g;
40
41	# split into min_len part and rest
42	my ($part,$rest) = ( substr($_,0,$min_len), substr($_,$min_len) );
43
44	# make part lowercase
45	$part = lc($part);
46
47	$last_part = $part if (! $last_part);
48
49	# new part?
50	if ($part ne $last_part) {
51	print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug);
52	$max_elements = $#part_arr if ($#part_arr > $max_elements);
53	print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
54	$total += $#part_arr;
55	@part_arr = ();
56	$last_part = $part;
57	}
58	push @part_arr, "'$_'";
59
60	# break out?
61	last if ($max && $total > $max);
62	}
63
64	print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr);
65	print "var min_len = $min_len;\n";
66	print "// index elements: $total\n";
67
68	print STDERR "You have more than $increase_at elements in one array,\nyou whould probably increase min_len to ",$min_len+1," or higher.\n" if ($max_elements > $increase_at);