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;

        my ($path,$headline) = split(/\t+/,$_,2);

        # escape single quote
        $headline =~ s/'/\\'/g;

        # split into min_len part and rest
        my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$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 && $#part_arr > $increase_at);
                $max_elements = $#part_arr if ($#part_arr > $max_elements);
                print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr);
                $total += $#part_arr;
                @part_arr = ();
                $last_part = $part;
        }
        push @part_arr, "['$path','$headline']";

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

print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr);
print qq{

var min_len = $min_len;
var html_pre = '<br/><a href="../';
var html_mid = '">';
var html_post = '</a>';

// index elements: $total

};

print STDERR "You have more than $increase_at elements, so you should\nincrease min_len to ",$min_len+1," or higher for performance benefit.\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 $max_elements = 0;
34
35	while(<STDIN>) {
36	chomp;
37
38	my ($path,$headline) = split(/\t+/,$_,2);
39
40	# escape single quote
41	$headline =~ s/'/\\'/g;
42
43	# split into min_len part and rest
44	my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) );
45
46	# make part lowercase
47	$part = lc($part);
48
49	$last_part = $part if (! $last_part);
50
51	# new part?
52	if ($part ne $last_part) {
53	print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug && $#part_arr > $increase_at);
54	$max_elements = $#part_arr if ($#part_arr > $max_elements);
55	print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr);
56	$total += $#part_arr;
57	@part_arr = ();
58	$last_part = $part;
59	}
60	push @part_arr, "['$path','$headline']";
61
62	# break out?
63	last if ($max && $total > $max);
64	}
65
66	print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr);
67	print qq{
68
69	var min_len = $min_len;
70	var html_pre = '<br/><a href="../';
71	var html_mid = '">';
72	var html_post = '</a>';
73
74	// index elements: $total
75
76	};
77
78	print STDERR "You have more than $increase_at elements, so you should\nincrease min_len to ",$min_len+1," or higher for performance benefit.\n" if ($max_elements > $increase_at);