35 |
while(<STDIN>) { |
while(<STDIN>) { |
36 |
chomp; |
chomp; |
37 |
|
|
38 |
|
my ($path,$headline) = split(/\t+/,$_,2); |
39 |
|
|
40 |
# escape single quote |
# escape single quote |
41 |
s/'/\\'/g; |
$headline =~ s/'/\\'/g; |
42 |
|
|
43 |
# split into min_len part and rest |
# split into min_len part and rest |
44 |
my ($part,$rest) = ( substr($_,0,$min_len), substr($_,$min_len) ); |
my ($part,$rest) = ( substr($headline,0,$min_len), substr($headline,$min_len) ); |
45 |
|
|
46 |
# make part lowercase |
# make part lowercase |
47 |
$part = lc($part); |
$part = lc($part); |
50 |
|
|
51 |
# new part? |
# new part? |
52 |
if ($part ne $last_part) { |
if ($part ne $last_part) { |
53 |
print STDERR $last_part,"\t",$#part_arr+1,"\n" if ($debug); |
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); |
$max_elements = $#part_arr if ($#part_arr > $max_elements); |
55 |
print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr); |
print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
56 |
$total += $#part_arr; |
$total += $#part_arr; |
57 |
@part_arr = (); |
@part_arr = (); |
58 |
$last_part = $part; |
$last_part = $part; |
59 |
} |
} |
60 |
push @part_arr, "'$_'"; |
push @part_arr, "['$path','$headline']"; |
61 |
|
|
62 |
# break out? |
# break out? |
63 |
last if ($max && $total > $max); |
last if ($max && $total > $max); |
64 |
} |
} |
65 |
|
|
66 |
print "headlines['$last_part'] = Array(\n ",join(",\n ",@part_arr),");\n" if (@part_arr); |
print "headlines['$last_part'] = [\n ",join(",\n ",@part_arr),"];\n" if (@part_arr); |
67 |
print "var min_len = $min_len;\n"; |
print qq{ |
68 |
print "// index elements: $total\n"; |
|
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 $increase_at elements in one array,\nyou whould probably increase min_len to ",$min_len+1," or higher.\n" if ($max_elements > $increase_at); |
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); |