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 |
25 |
my $min_len = shift @ARGV || 3; |
11 |
dpavlin |
4 |
# if more than x elements, warn to increase min_len |
12 |
|
|
my $increase_at = 500; |
13 |
dpavlin |
1 |
|
14 |
dpavlin |
10 |
# name of generated index |
15 |
|
|
my $headlines = 'headlines'; |
16 |
|
|
|
17 |
dpavlin |
3 |
my $debug = 1; |
18 |
|
|
|
19 |
dpavlin |
1 |
sub print_file { |
20 |
|
|
my $f = shift || return; |
21 |
|
|
open(F, $f) || die "$f: $!"; |
22 |
|
|
while(<F>) { |
23 |
|
|
print; |
24 |
|
|
} |
25 |
|
|
close(F); |
26 |
|
|
} |
27 |
|
|
|
28 |
|
|
print qq{ |
29 |
dpavlin |
10 |
var $headlines = new Object(); |
30 |
dpavlin |
1 |
}; |
31 |
|
|
|
32 |
|
|
my @part_arr; |
33 |
|
|
my $last_part = ''; |
34 |
|
|
my $total = 0; |
35 |
|
|
|
36 |
dpavlin |
4 |
my $max_elements = 0; |
37 |
|
|
|
38 |
dpavlin |
25 |
sub escape_js { |
39 |
|
|
my $t = shift || return; |
40 |
|
|
$t =~ s/(['\\])/\\$1/g && print STDERR "ESCAPED '$t'\n"; |
41 |
|
|
return $t; |
42 |
|
|
} |
43 |
|
|
|
44 |
dpavlin |
1 |
while(<STDIN>) { |
45 |
|
|
chomp; |
46 |
|
|
|
47 |
dpavlin |
9 |
if (!m/\t/ || m/\t$/) { |
48 |
|
|
print STDERR "SKIP '$_': no tab\n"; |
49 |
|
|
next; |
50 |
|
|
} |
51 |
|
|
|
52 |
dpavlin |
25 |
my @data = split(/\t+/,$_); |
53 |
dpavlin |
7 |
|
54 |
dpavlin |
25 |
my $headline = shift @data || die "need at least headline!"; |
55 |
|
|
|
56 |
dpavlin |
9 |
if (length($headline) < $min_len) { |
57 |
|
|
print STDERR "SKIP '$_': too short\n"; |
58 |
|
|
next; |
59 |
|
|
} |
60 |
dpavlin |
1 |
|
61 |
dpavlin |
9 |
|
62 |
dpavlin |
1 |
# split into min_len part and rest |
63 |
dpavlin |
7 |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
64 |
dpavlin |
1 |
|
65 |
|
|
# make part lowercase |
66 |
|
|
$part = lc($part); |
67 |
|
|
|
68 |
|
|
$last_part = $part if (! $last_part); |
69 |
|
|
|
70 |
|
|
# new part? |
71 |
|
|
if ($part ne $last_part) { |
72 |
dpavlin |
7 |
print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug && $#part_arr > $increase_at); |
73 |
dpavlin |
4 |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
74 |
dpavlin |
25 |
print "${headlines}['",escape_js($last_part),"'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
75 |
dpavlin |
1 |
$total += $#part_arr; |
76 |
|
|
@part_arr = (); |
77 |
|
|
$last_part = $part; |
78 |
|
|
} |
79 |
dpavlin |
25 |
push @part_arr, "['".escape_js($headline)."','".join("','",map { escape_js($_) } @data)."']"; |
80 |
dpavlin |
1 |
|
81 |
|
|
# break out? |
82 |
|
|
last if ($max && $total > $max); |
83 |
|
|
} |
84 |
|
|
|
85 |
dpavlin |
25 |
print "${headlines}['",escape_js($last_part)."'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
86 |
dpavlin |
7 |
print qq{ |
87 |
dpavlin |
4 |
|
88 |
dpavlin |
10 |
${headlines}.min_len = $min_len; |
89 |
|
|
${headlines}.length = $total; |
90 |
dpavlin |
7 |
|
91 |
|
|
}; |
92 |
|
|
|
93 |
|
|
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); |