7 |
# maximum entries |
# maximum entries |
8 |
my $max = 0; |
my $max = 0; |
9 |
# minimum letters to search by |
# minimum letters to search by |
10 |
my $min_len = 3; |
my $min_len = shift @ARGV || 3; |
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 |
|
|
35 |
|
|
36 |
my $max_elements = 0; |
my $max_elements = 0; |
37 |
|
|
38 |
|
sub escape_js { |
39 |
|
my $t = shift || return; |
40 |
|
$t =~ s/(['\\])/\\$1/g && print STDERR "ESCAPED '$t'\n"; |
41 |
|
return $t; |
42 |
|
} |
43 |
|
|
44 |
while(<STDIN>) { |
while(<STDIN>) { |
45 |
chomp; |
chomp; |
46 |
|
|
49 |
next; |
next; |
50 |
} |
} |
51 |
|
|
52 |
my ($path,$headline) = split(/\t+/,$_,2); |
my @data = split(/\t+/,$_); |
53 |
|
|
54 |
|
my $headline = shift @data || die "need at least headline!"; |
55 |
|
|
56 |
if (length($headline) < $min_len) { |
if (length($headline) < $min_len) { |
57 |
print STDERR "SKIP '$_': too short\n"; |
print STDERR "SKIP '$_': too short\n"; |
62 |
# split into min_len part and rest |
# split into min_len part and rest |
63 |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
64 |
|
|
|
# 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; |
|
|
|
|
65 |
# make part lowercase |
# make part lowercase |
66 |
$part = lc($part); |
$part = lc($part); |
67 |
|
|
71 |
if ($part ne $last_part) { |
if ($part ne $last_part) { |
72 |
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); |
73 |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
74 |
print "${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); |
75 |
$total += $#part_arr; |
$total += $#part_arr; |
76 |
@part_arr = (); |
@part_arr = (); |
77 |
$last_part = $part; |
$last_part = $part; |
78 |
} |
} |
79 |
push @part_arr, "['$path','$headline']"; |
push @part_arr, "['".escape_js($headline)."','".join("','",map { escape_js($_) } @data)."']"; |
80 |
|
|
81 |
# break out? |
# break out? |
82 |
last if ($max && $total > $max); |
last if ($max && $total > $max); |
83 |
} |
} |
84 |
|
|
85 |
print "${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); |
86 |
print qq{ |
print qq{ |
87 |
|
|
88 |
${headlines}.min_len = $min_len; |
${headlines}.min_len = $min_len; |