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 |
29 |
my $min_len = shift @ARGV; |
11 |
|
|
$min_len = 3 unless defined($min_len); |
12 |
dpavlin |
4 |
# if more than x elements, warn to increase min_len |
13 |
|
|
my $increase_at = 500; |
14 |
dpavlin |
1 |
|
15 |
dpavlin |
10 |
# name of generated index |
16 |
|
|
my $headlines = 'headlines'; |
17 |
|
|
|
18 |
dpavlin |
3 |
my $debug = 1; |
19 |
|
|
|
20 |
dpavlin |
1 |
sub print_file { |
21 |
|
|
my $f = shift || return; |
22 |
|
|
open(F, $f) || die "$f: $!"; |
23 |
|
|
while(<F>) { |
24 |
|
|
print; |
25 |
|
|
} |
26 |
|
|
close(F); |
27 |
|
|
} |
28 |
|
|
|
29 |
|
|
print qq{ |
30 |
dpavlin |
10 |
var $headlines = new Object(); |
31 |
dpavlin |
1 |
}; |
32 |
|
|
|
33 |
|
|
my @part_arr; |
34 |
|
|
my $last_part = ''; |
35 |
|
|
my $total = 0; |
36 |
|
|
|
37 |
dpavlin |
4 |
my $max_elements = 0; |
38 |
|
|
|
39 |
dpavlin |
25 |
sub escape_js { |
40 |
dpavlin |
26 |
my $t = shift || return 'undef'; |
41 |
|
|
# escape single quote and backspace |
42 |
dpavlin |
25 |
$t =~ s/(['\\])/\\$1/g && print STDERR "ESCAPED '$t'\n"; |
43 |
dpavlin |
26 |
# quote string if not number |
44 |
|
|
$t = "'$t'" unless ($t =~ m/^\d+$/); |
45 |
dpavlin |
25 |
return $t; |
46 |
|
|
} |
47 |
|
|
|
48 |
dpavlin |
1 |
while(<STDIN>) { |
49 |
|
|
chomp; |
50 |
|
|
|
51 |
dpavlin |
9 |
if (!m/\t/ || m/\t$/) { |
52 |
|
|
print STDERR "SKIP '$_': no tab\n"; |
53 |
|
|
next; |
54 |
|
|
} |
55 |
|
|
|
56 |
dpavlin |
25 |
my @data = split(/\t+/,$_); |
57 |
dpavlin |
7 |
|
58 |
dpavlin |
25 |
my $headline = shift @data || die "need at least headline!"; |
59 |
|
|
|
60 |
dpavlin |
9 |
if (length($headline) < $min_len) { |
61 |
|
|
print STDERR "SKIP '$_': too short\n"; |
62 |
|
|
next; |
63 |
|
|
} |
64 |
dpavlin |
1 |
|
65 |
dpavlin |
9 |
|
66 |
dpavlin |
1 |
# split into min_len part and rest |
67 |
dpavlin |
7 |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
68 |
dpavlin |
1 |
|
69 |
|
|
# make part lowercase |
70 |
|
|
$part = lc($part); |
71 |
|
|
|
72 |
|
|
$last_part = $part if (! $last_part); |
73 |
|
|
|
74 |
|
|
# new part? |
75 |
|
|
if ($part ne $last_part) { |
76 |
dpavlin |
7 |
print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug && $#part_arr > $increase_at); |
77 |
dpavlin |
4 |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
78 |
dpavlin |
26 |
print "${headlines}[",escape_js($last_part),"] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
79 |
dpavlin |
1 |
$total += $#part_arr; |
80 |
|
|
@part_arr = (); |
81 |
|
|
$last_part = $part; |
82 |
|
|
} |
83 |
dpavlin |
26 |
push @part_arr, "[".escape_js($headline).",".join(",",map { escape_js($_) } @data)."]"; |
84 |
dpavlin |
1 |
|
85 |
|
|
# break out? |
86 |
|
|
last if ($max && $total > $max); |
87 |
|
|
} |
88 |
|
|
|
89 |
dpavlin |
26 |
print "${headlines}[",escape_js($last_part)."] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
90 |
dpavlin |
7 |
print qq{ |
91 |
dpavlin |
4 |
|
92 |
dpavlin |
10 |
${headlines}.min_len = $min_len; |
93 |
|
|
${headlines}.length = $total; |
94 |
dpavlin |
7 |
|
95 |
|
|
}; |
96 |
|
|
|
97 |
|
|
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); |