2 |
# |
# |
3 |
|
|
4 |
use strict; |
use strict; |
|
use locale; |
|
5 |
|
|
6 |
# maximum entries |
# maximum entries |
7 |
my $max = 0; |
my $max = 0; |
8 |
# minimum letters to search by |
# minimum letters to search by |
9 |
my $min_len = 3; |
my $min_len = shift @ARGV; |
10 |
|
$min_len = 3 unless defined($min_len); |
11 |
# if more than x elements, warn to increase min_len |
# if more than x elements, warn to increase min_len |
12 |
my $increase_at = 500; |
my $increase_at = 500; |
13 |
|
|
14 |
|
# name of generated index |
15 |
|
my $headlines = 'headlines'; |
16 |
|
|
17 |
my $debug = 1; |
my $debug = 1; |
18 |
|
|
19 |
sub print_file { |
sub print_file { |
26 |
} |
} |
27 |
|
|
28 |
print qq{ |
print qq{ |
29 |
var headlines = Array(); |
var $headlines = new Object(); |
30 |
}; |
}; |
31 |
|
|
32 |
my @part_arr; |
my @part_arr; |
35 |
|
|
36 |
my $max_elements = 0; |
my $max_elements = 0; |
37 |
|
|
38 |
|
sub escape_js { |
39 |
|
my $t = shift || return 'undef'; |
40 |
|
# escape single quote and backspace |
41 |
|
$t =~ s/(['\\])/\\$1/g && print STDERR "ESCAPED '$t'\n"; |
42 |
|
# quote string if not number |
43 |
|
$t = "'$t'" unless ($t =~ m/^\d+$/); |
44 |
|
return $t; |
45 |
|
} |
46 |
|
|
47 |
|
my @lines; |
48 |
|
|
49 |
while(<STDIN>) { |
while(<STDIN>) { |
50 |
chomp; |
chomp; |
51 |
|
|
54 |
next; |
next; |
55 |
} |
} |
56 |
|
|
57 |
my ($path,$headline) = split(/\t+/,$_,2); |
# remove leading spaces (which are ignored if source list was |
58 |
|
# sorted using locale) |
59 |
|
s/^\s+//; |
60 |
|
|
61 |
|
push @lines, $_; |
62 |
|
} |
63 |
|
|
64 |
|
foreach (sort { lc($a) cmp lc($b) } @lines) { |
65 |
|
|
66 |
|
my @data = split(/\t+/,$_); |
67 |
|
|
68 |
|
my $headline = shift @data || die "need at least headline!"; |
69 |
|
|
70 |
if (length($headline) < $min_len) { |
if (length($headline) < $min_len) { |
71 |
print STDERR "SKIP '$_': too short\n"; |
print STDERR "SKIP '$_': too short\n"; |
76 |
# split into min_len part and rest |
# split into min_len part and rest |
77 |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
78 |
|
|
|
# escape special chars |
|
|
$part =~ s/(['\\])/\\$1/g && print STDERR "ESCAPED part '$part'\n"; |
|
|
$rest =~ s/(['\\])/\\$1/g && print STDERR "ESCAPED rest '$rest'\n"; |
|
|
$headline =~ s/(['\\])/\\$1/g; |
|
|
|
|
79 |
# make part lowercase |
# make part lowercase |
80 |
$part = lc($part); |
$part = lc($part); |
81 |
|
|
85 |
if ($part ne $last_part) { |
if ($part ne $last_part) { |
86 |
print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug && $#part_arr > $increase_at); |
print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug && $#part_arr > $increase_at); |
87 |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
88 |
print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
print "${headlines}[",escape_js($last_part),"] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
89 |
$total += $#part_arr; |
$total += $#part_arr; |
90 |
@part_arr = (); |
@part_arr = (); |
91 |
$last_part = $part; |
$last_part = $part; |
92 |
} |
} |
93 |
push @part_arr, "['$path','$headline']"; |
push @part_arr, "[".escape_js($headline).",".join(",",map { escape_js($_) } @data)."]"; |
94 |
|
|
95 |
# break out? |
# break out? |
96 |
last if ($max && $total > $max); |
last if ($max && $total > $max); |
97 |
} |
} |
98 |
|
|
99 |
print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
print "${headlines}[",escape_js($last_part)."] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
100 |
print qq{ |
print qq{ |
101 |
|
|
102 |
var min_len = $min_len; |
${headlines}.min_len = $min_len; |
103 |
var html_pre = '<div><a href="../'; |
${headlines}.length = $total; |
|
var html_mid = '">'; |
|
|
var html_post = '</a></div>'; |
|
|
|
|
|
// index elements: $total |
|
104 |
|
|
105 |
}; |
}; |
106 |
|
|